Aspect, Evidentiality and Tense in Mongolian from Middle Mongol to Khalkha and Khorchin
Aspect, evidentiality and tense in Mongolian From Middle Mongol to Khalkha and Khorchin
Benjamin Brosig ©Benjamin Brosig, Stockholm University 2014
ISB N 978-91-628-9124-4
Printed in Sweden by US-AB, Stockholm 2014 Distributor: Department of linguistics, Stockholm University
ii Ȼɚɹɱɭɭɞɢɣɧɯɚɚɷɪɯɦɷɞɥɷɷɫɝɚɪɫɚɧ Ȼԛɯɹɫɬɧɵɯɚɚ ɫɨɺɥɢɣɝɧɷɝɬɝɷɫɷɧ ɍɫɵɝ ɧɶ ɭɭɠ ɺɫɵɝ ɧɶ ɞɚɝɚɚɝԛɣ Ԛɧɞɫɷɧԛɡɥɷɷ ɨɪɯɢɠɭɯɚɚɧɚɚɨɥɫɨɧ ɗɪɢɣɧ ɞɷɷɪɷɧɝԛɣɱɚɧɚɪɚɚɯɚɹɫɚɧ ɗɪɯ ɬɷɝɲ ɚɦɶɞɪɚɥɬɚɣ , ɷɝɷɥ ɯɚɣɪɥɚ ɯɚɚɪ ɦɨɧɝɨɥɧɢɣɝ ɦɢɣɧ ɬԧɥԧԧ
iii iv Abstract
The present thesis consists of an introduction and the following papers:
x The aspect-evidentiality system of Middle Mongol. Ural-Altaic Studies, 13 . (forthcoming) x The tense-aspect system of Khorchin Mongolian. In: Pirkko Suihkonen & Lindsay Whaley (eds.), Typology of Languages of Europe and Northern and Central Asia . Amsterdam: John Benjamins. (forthcoming) x Aspect and epistemic notions in the present tense system of Khalkha Mongoli- an. Acta Linguistica Petropolitana . (forthcoming) x Factual vs. evidential? - The past tense forms of spoken Khalkha Mongolian. In: Ad Foolen, Helen de Hoop, & Gijs Mulder (eds.), Empirical Approaches to Evidentiality . Amsterdam: John Benjamins. (under review)
Its purpose is to give an account of tense, aspect and evidentiality in three Mongolian varieties: Middle Mongol (MM) as spoken in the Mongol Em- pire, Khalkha Mongolian as spoken in the Mongolian state, and Khorchin Mongolian as spoken in eastern Inner Mongolia, China. MM started out with a tripartite tense distinction and a medium-sized aspectual system. Its past evidential system was tripartite with suffixes for firsthand, non-firsthand and evidentially neutral information. In Khorchin, which developed under the influence of Mandarin and Manchu, evidentiality was lost, and tense was simplified into a past / non-past distinction, alongside with a discontinuous proximal future / past marker. The aspect system underwent some changes, but retained its complexity. Khalkha, which developed under the influence of Turkic and Tibetan, underwent some shared innovations with Khorchin, but retained participles as a multifunctional unit within finite predicates, so that its aspectual system grew more complex. The past evidentiality distinc- tions of MM were basically retained, but the introduction of present tense evidentiality brought a number of changes: the evidentially neutral value shifted to signaling assimilated knowledge, and discontinuous future uses were introduced for all past markers.
v vi Contents
Abstract ...... v Acknowledgements ...... xi
ʡĄʒĄ˃ǶĄʓ ...... xv Introduction ...... 1 1. Classification of Mongolic ...... 2 2. Terminology ...... 4 2.1 Tense ...... 4 2.2 Aspectuality ...... 5 2.3 Evidentiality ...... 8 2.4 Structural grammatical terms ...... 9 3. Previous research and methodology ...... 9 4. The case studies of TAE in Mongolian ...... 12 4.1 The aspect-evidentiality system of Middle Mongol ...... 13 4.2 The tense-aspect system of Khorchin Mongolian ...... 14 4.3 Aspect and epistemic notions in the present tense system of Khalkha Mongolian ...... 15 4.4 Factual vs. evidential? - The past tense forms of spoken Khalkha Mongolian ...... 16 5. From Middle Mongol to Central Mongolic ...... 17 5.1 Finitely usable past forms ...... 17 5.2 Finitely usable present forms ...... 20 5.3 Aspectual constructions in finite predicates ...... 23 5.4 Possible combinations ...... 24 5.5 Evidential particles ...... 26 6. No proof required: Khorchin in Western Manchuria ...... 26 7. Mixed evidentiality: Khalkha between Central Asia and Amdo ...... 33 8. Closing words ...... 40 8.1 Evidentiality in Khalkha: a synthesis ...... 40 8.2 Perspectives for areal studies on TAE in Greater Mongolia ...... 43 8.3 A short note on the classification of Central Mongolic ...... 44 Bibliography ...... 45 Aspect, tense and evidentiality in Middle Mongol ...... 52 1. Inventory ...... 54 vii 2. Corpus ...... 58 3. Imperfective domain ...... 58 3.1 The generic -yU and the progressive-habitual -mU ...... 59 3.2 The present progressive -n bu-yu ...... 68 3.3 Past imperfective -n a-ju ѫu / bü-lüge ...... 69 3.4 Future -QU and the form -n a-qu ...... 71 3.5 Future in the past -QU bü-lüge / a-ju ѫu, and -QU bol-...... 75 3.6 -d ...... 79 4. Perfective domain ...... 81 4.1 Perfective past -bA ,-lUGA and -jUGU ...... 81 4.2 Resultative-Progressive -jU a-ju ѫu / bü-lüge / a-mu ...... 89 4.3 Perfect -GsAn (a-ju ѫu / bü-lüge ) ...... 91 4.4 -GA ,-GA bü-lüge ...... 97 5. Conclusion ...... 99 Bibliography ...... 102 The tense-aspect system of Khorchin Mongolian ...... 105 1. Introduction ...... 105 2. Preliminaries ...... 106 3. Data collection and research methdology ...... 107 4. Structure of the Khorchin TA system ...... 109 5. Markers of the Khorchin TA system ...... 112 5.1 Central system ...... 113 5.2 Minor simple forms and their combinability ...... 140 5.3 Non-verbal predications ...... 148 5.4 Non-temporal and non-aspectual notions ...... 150 6. Areal and diachronic factors: a preliminary evaluation ...... 163 Glossings ...... 168 Bibliography ...... 168
Aspect and epistemic notions in the present tense system of Khalkha Mongolian ...... 171 1. Corpus, informants and evaluation ...... 172 2. Inventory ...... 173 2.1 Inventory of forms ...... 173 2.2 Forms and their meanings ...... 175 3. The evidential forms -n and -aa ...... 175 3.1 Simple -n ...... 176 3.2 Simple -aa ...... 185 3.3 -n and -aa at the end of complex forms ...... 188 4. The Habitual ...... 206 4.1 Habitual use of simple -dag ...... 206 4.2 Mono-occasional use of simple -dag ...... 209 4.3 -dag at the end of complex forms ...... 212
viii 5. -h...... 218 5.1 -h with particles / clitics relativizing factuality ...... 219 5.2 -h followed by ge - or in absolute final position ...... 221 5.3 -h in complex forms with the copula...... 223 6. -aad ...... 224 7. = l...... 228 8. Conclusion ...... 231 Bibliography ...... 234
Factual vs. evidential? - The past tense forms of spoken Khalkha Mongolian ...... 237 1. Introduction ...... 237 2. Linguistic evidence ...... 239 3. Frequencies ...... 240 4. Established Past -sang ...... 241 5. Firsthand Past -laa ...... 245 6. Non-Firsthand Past -jee ...... 248 7. -v ...... 251 8. Questions ...... 252 9. Futures ...... 254 10. Conclusions ...... 257 Bibliography ...... 259 Swedish summary ...... 262
ix x Acknowledgements
First and foremost, in order to design and implement the research of this dissertation, a great deal of knowledge and training in fields such as tense, aspect, modality and evidentiality, linguistic typology, pragmatics, linguistic evidence, field linguistics, Mongolian conversation, Mongolian dialect stud- ies, historical linguistics and “Altaic” studies was necessary. It was mainly my study at institutions such as the former department of Central Asian stud- ies in Bonn with teachers such as Stefan Georg, Senderjaviin Alimaa, Rudolf Kaschewsky and Oyun, the department of linguistics in Cologne with Hans- Jürgen Sasse, Leila Behrens and Dagmar Jung, and the department of lin- guistics in Stockholm with Östen Dahl, Maria Koptjevskaja Tamm and all the others that provided me with this knowledge. Other places that contribut- ed to the intellectual foundations of this thesis were the department of Japa- nese studies and the center for East Asian languages in Bonn, the School of Oriental and African Studies in London, and the department of Mongolian studies at the Inner Mongolian University in Höhhot. Then, there were many individuals from whose input I benefitted. Östen as my main supervisor was good to work with, requiring precise thought and its exact expression, challenging one-sided interpretations of secondary sources, and providing quick feedback when needed. Masha, my second supervisor, usually had a keen sense of structure, organization, and practical, implementable solutions. They forced me to not only write a good disserta- tion - something I had intended to do anyway - but a moderately readable one, teaching me to empathize with my readers in the course. John Street (Wisconsin-Madison), whom I definitely counted in my target audience, also played a part in this process. Moreover, he contributed a number of valuable comments - and, for that matter, corrections! - to my analysis of Middle Mongol. For Khalkha Mongolian, D. Guntsetseg (Stuttgart / München) proved to be a reliable shortcut for understanding difficult examples. She also prevented several diffuse (not TAME-related) translation mistakes, and her professional assessments led me in some cases to deemphasize outliers in my elicitation results. In the case of non-habitual -dag, her suggestions ena- bled me to sharpen my analysis. For Khorchin, it was a wide, but more light- ly concentrated network of people mainly from Inner Mongolian Nationali- ties University (ÖMÜYS), Central Nationalities University (TÜYS) and Inner Mongolian University (ÖMYS) who supported me. Before the others, I must mention B. Sodu and his student Engkebayar (ÖMÜYS) without whose xi thoughtful, timely support my second field trip to Tongliao might well have ended in failure. During my third fieldtrip, I could have gotten somewhat improved results and - most of all - avoided a lot of frustration if I had un- derstood the need to accept the help that Sodu had generously offered. In the field, it was Guo Ba ܵatur, Qous, Xiaohua, Köke-Aܵula, Yunguan, Sarana and a few others that helped me to collect data and get into contact with in- formants. Naturally, transcriptions of the collected materials were of central importance. A number of people contributed here, but most work was done by two people. For Khorchin, the trained phonetician Han Guojun (then ÖMYS) provided me with a phonemic transcription of a large part of my Khorchin materials as reliable as anybody could produce, defying dialectal problems that he as a native of the Eastern Back Banner encountered with materials from the Eastern Central Banner. For Khalkha, my dear friend B. Zoljargal, a trained ethnologist from Ulaanbaatar, almost single-handedly transcribed huge amounts of data. Regrettably, it was not possible to use Robert Östling’s automated annotations of the Khalkha corpus for my thesis, but I am grateful for his work, which will no doubt be useful for me at a later point in time. Among the other scientists who at some point or the other supported me, I want to thank %ULQVDLQ .\ǀWǀ Urancimeg (Cambridge), Oyungerel and Taibung (ÖMYS) for their advice concerning Khorchin history, society and the literature on it (though I only had opportunity to use a very small part of the latter), Duran, Bayanco ܵtu (ÖMYS), Soduba ܵatur, Orulamjab, the musi- cologist Gow-a (TÜYS) and my pre-opponent Henrik Bergqvist for their more linguistically oriented advice, To ܵtambayar (Inner Mongolian Poly- technic), M. Bayarsaihan (Mongolian National University = MUIS), C. Gantulga (Ulaanbaatar), Éva Csató (Uppsala), Christopher Atwood (Indi- ana), Borjigin Manglai (TÜYS), Abdurishid Yakup (Berlin), Buyanba ܵatur (Inner Mongolia Normal University = ÖMBYS), Volker Rybatzki (Helsin- ki), Britta-Maria Gruber (Bonn) and, most of all, Ilja Gruntov (Moscow) for suggesting and providing me with literature, and Gow-a (ÖMYS), Carsten Friede (Bonn), Galsang, Tul ܵaܵuri (ÖMÜYS), Erdenimöngke (ÖMBYS), .|NHEDܵDWXU 6KRZD :RPHQތV 8QLYHUVLW\ , Köke (TÜYS), J. Bayansan (MUIS) and especially Kürelba ܵatur (Cambridge / ÖMBYS) and Qascimeg (TÜYS) for their help in organizational issues. Particular thanks go to Se- cenba ܵatur (ÖMYS) and Secenco ܵtu (TÜYS) for their advice and support from 2006 up to now. My thesis would doubtlessly look very different from now if all the un- named informants and also a number of people I chose not to single out or simply forgot to mention had not been there. Another category of people missing are those who contributed to linguistic endeavors that turned out to be distinct from this thesis, but will hopefully turn into something later. My heartfelt thanks go to all who contributed to making this thesis possible, and
xii I hope that the existence of this thesis or what it might mean to them may compensate them for their efforts. At Stockholm University, the department of linguistics, with its typolo- gists, field researchers, computational linguists and sign language research- ers, fikas and fruits, provided a functioning, pleasant and inspiring research environment. Computational linguistics blended in very well, and the contact with sign language studies, an area that ought to be much better integrated into linguistics and is relevant for both field linguistics and the understand- ing of human linguistic cognition, was an important experience. Considering the human conditions that allowed me to complete this the- sis, one would first have to name some of my fellow doctoral students: my office mates Thomas [Hörberg], Pia [Simper-Allen], Desu [Desalegn Hagos] and Calle [Börstell], Emil [Perder], Robert [Östling], and, above all, Yvi [Yvonne Agbetsoamedo] (who can now swim quite veritably) and Fra [Francesca di Garbo]. And Tögsöö. Even living space was temporalily over- lapping, with Emil, Yvonnne, Marjatta Sikström and Niklas Öhrström. Without the fortress of Europe, progressing social deforestation and a local bourgeois that exerted great control over the forms of accommodation avail- able, not to mention without an imprisoning sense of responsibility, Stock- holm might indeed have been a splendid place to live. During my various stays in Ulaanbaatar, it was friends like Alimaa [W.-H.], Enhee [Enhtuyaa], Hishii [Hishigjargal] and Gantulga that made my life worthwhile. In 2011, I spent the most beautiful summer of my life with Tergel [Zolzayaa]. Most of all, I am grateful to my old friend Zoloo [Zoljargal], a person whose non- existence would have left a void somewhere. While my research plans gave me little opportunity to stay in Beijing for longer, my contact with Miáomiáo meant and means a lot to me. So do /LԃhQHQFLEDܵDWXU@ 6DLKQD a [Batsaihan], Byambaa [Byambasüren], Zandaa [Zandanhüü], Ma Fuchuan, Zhao Secen and Orulamjab. Not to forget my sister Larissa, my parents Werner and Petra, my four grandparents, who all lived to witness the completion of my thesis, and Meiki, who didn’t. And there is Orgil, the only person without whom this thesis, in any form, would definitely never have been written. 2
1嘥↛ᅉ◊Ꮫᡤ⣼㸪汄᭷ᮘᅾி攧ஂ㏸␃㸪ణⱑⱑⓗ 㸪彯ཤ 䍘ᅾ㸪㒔⮡ᡃᘺ ㊊⌋ 峝ࠋี䂐*HUOHH ஓ 征㟟 ࠋᅾᡃ孢ᩥ⟅ 彑ᮇ斜ᡃẔἐ᭷ொᮘ᥋ゐࠋ 2 Was die menschlichen Umstände angeht, die es mir erlaubt haben, meine Doktorarbeit zu schreiben, so sollte ich zuerst einmal meine Mitdoktoranden erwähnen: Thomas Hörberg, Pia Simper-Allen, Desalen Hagoss und Calle Börstell, welche mit mir ein Büro teilen, Emil Per- der, Robert Östling und, allen voran, „Iwi“ (Iwonn Agbätscho-Amedo) und Fra(nschi) (Frant- schäska di Garbo). Des Weiteren Tugßo. Selbst der Wohnraum überlappte kurzzeitig mit Leuten von meiner Universität, nämlich mit Emil, Iwonn, Marjatta Sikström und Niklas xiii During the entire time of my doctoral studies, I was employed by Stock- holm University with a full salary and social benefits - appropriate research conditions that are usually withheld from doctoral students in Germany. In addition, the department of linguistics financed one and a half of my field trips and enabled me to attend (around) five conferences. The costs of a sev- en-month stay at the National University of Mongolia in Ulaanbaatar during 2013 were covered by a grant of Sven och Dagmar Saléns stiftelse. In addi- tion, I received traveling grants for conferences from Knut och Alice Wal- lenbergs Stiftelse (Höhhot 2012), Kinanders stiftelse (Göttingen 2014) and Gålöstiftelsen (Groningen 2014).
Öhrström. Ohne die Festung Europa, fortschreitenden Sozialabbau und eine örtliche Bour- geoise, die einen massiven negativen Einfluss auf die Wohnmöglichkeiten ausübt, und nicht zu vergessen ohne ein zum Gefängnis taugendes Verantwortungsgefühl hätte Stockholm durchaus ein angenehmer Wohnort sein können. Während meiner mehrfachen Aufenthalte in Ulaanbaatar waren es Freunde wie Alima, Enche [Enchtoja], Hiishii und Gantolag, die das Leben lebenswert machten. 2011 verbrachte ich den schönsten Sommer meines Lebens zu- sammen mit Tirrgill. Allen anderen voran bin ich meiner alten Freundin Dsolloo dankbar, ein Mensch, deren Nichtexistenz bei mir eine Leere hinterlassen müsste. Während meine For- schungspläne mir kaum Gelegenheit gaben, mich längere Zeit in Peking aufzuhalten, so be- deutete und bedeutet mir meine Beziehung zu Mjaumjau doch viel. Dasselbe gilt für Lju Je und Girrlee, die ich während der ganzen Zeit in Stockholm fast gar nicht treffen konnte. Wei- tere Leute, mit denen ich zwar insgesamt weniger zu tun hatte, aber die zu einem gegebenen Zeitpunkt einen wichtigen Unterschied gemacht haben, sind die beiden Gowa-s, Bao Hongli, Lin-Ii, Baatar, ßäächnaa, Bjambaa, Dsandaa, Ma Futschüän, Dschao ßetschen und Orrlommd- schaf. Und nicht zu vergessen meine Schwester Larissa, meine beiden Eltern Werner und Petra, meine vier Großeltern, die alle den Zeitpunkt meiner Verteidigung noch erleben konn- ten, und Meiki, der das nicht vergönnt war. Und schließlich ist da noch Orrgill, die einzige Person, ohne die diese Doktorarbeit ganz sicher niemals und in keinerlei Form geschrieben worden wäre. xiv Ĭʣ˄ĬăĞ۩Ğő ŘʳŹĞ۩ĸ ĞĠŋʒʹʦ Ĵ˃ɐő ʋĞŎĠ۩ŋĕʒĞʹĕʦ ŹʯĞȟ۩ʫĸ ĝŘɐʳŀ ăʳĞʓ۩ĞĞĄő ʫĬ˺ʣĬ˄۩Ĭő Ƞ˃Ğʹĕő۩ĮʖĄɆĬ˃ĠʋĕʬĕʒĕɊ ʡĄʒĄ˃ǶĄʓ ɎĞő Ąè ĞʬʓűĞʳè ĕĕʒʓľēēɮʗƎ˃Ġ۩ĞĞő ēȁ ʋĕʬĕɐʒĕʓ۩ľő űĞĞʌʳĠè űĞĞʬĄʓè ʮĄȽè ļɎ۩ĞĞő ʳǵʒȿĄè ɐ۩۩ĞűĞĄĄʢè ĕȃȂĖ ʖĞōʳĞʒĕʓè ȂʒĖ ʡĕľȃő űĞĞʯĄɆĄʒʢćè Ȃɐ˄۩ė۩ĞĞő űĞ ǵĬʒƂȿʬĄʓè ēʳĕʌʗĞɌʗĕő ăʳĮ ă˃Ƀċ ʲĽĠ۩ĞĠ ʲĕ˃Ɏ ȂʒĖ ʸĄ˃ĞʹĄő۩ŀ ʋĬōɌɆĬʓ ʒĮ ĬĬĬè ĵăĠĞő ȂʒĖ ăĠ۩ĞĞő ʡĵŭ ʖĬʬĬʒĬʓè ăʹĄʒɆĮ ʋĕʬĕʒĕɊ۩ēʯĖʹ ɌĕűĞʗőʡĞʌ۩ʯēĖ ĕĕĕ۩ŋʬʬĴɌɮɌʗ ŋĄʬĄʬĮ ʋĕʬĕʒĕɊ۩Ġ ēŎĖ ʡĕĞĞʌŀ۩ēʯĖ űĞĞȿʗĄőé Ȃ˃ĕɌʢĕĠ˃ ʲĕ˃ɎăĠèȂŭʖĞōʳĠ۩ĞĞőȂʒĖʖĞōʳĞʒĕʓèȂ˃ĕɌʒĕɎèʋĬōɌɆĬʓȂ ĄĬ۩ʡĕ˃Ğɮő۩ʫĸ ʸĄɆĬő۩Į
ȾĄʬĄŔċ
ʋĵőȂʒĖʖĞŎĕʳĞʒĕʓ۩ĬőŋĬʢĄʒĄɃċèű˃Ğʌʢć۩ĞĞő
ōĄĄʧ˃ʨĮ Ľʓ Ŷè ēŎĖ ɏŶʓè ʲĽĞʓ ʧĬ˃ĄʨǶĮ īōʯĄʒĄő Ȃ˃ĕɌʢĕĠűĞĞȿʗĄő
ʖĬʬĬʒĬʓ ůʡĬǶĄĞĞʒĄŲʓ
ʲĕ˃Ɏ۩ʸĕ˃ĶŘȃĠ
ļĞʒĖ ĞǶĮ ʒĖ ĕő
ˍĽĬˍ ʝˇ̇īĬè ̔ʒő۩ŀ īʹĬőè ̀ʝňˇʗ̇è ˂ĽʬĬʒ ĬĬĬ۩ő ĕɘ۩ő ʣˑɻ˃è ňʬ˃Ą۩ő ăʒĞʍ۩Ĉè ʖňōʬĕ˃ʳĄŭ۩Ĭő Ɓő۩Į ɻĬ˃˻è ʖʣňˑő ʡĕŘɘʍ۩Ĭő ʖĬʬĬʒĬʓ۩Ĭő ĞʳʒʓľʡŘʍľͻōʸ˃ő Ąʗè ĄĞ۩ƕ˃ĕĕʘè ʑĄĞĞʓ۩Ĉ ʖĄʗʗʼnè ʸĽ˃ɼő ͻĄōʘ ʡĕŘȋʍ۩ľő ʖĞōʳĞʒĕʓ۩ľő Ğ۩ĬĬĬ۩ő ĕʬʖ˃ĄĬĠʋő ȃĬćĞő ʸĕȁ ȠȃǶĬʢć۩ĞĞő ʋĵő ʖĬ˃ɆĄɆĬʒĠè ʧĕɐʬŀ ăˌ˃Ğ˻۩ʖĬʬĬʒĬʓ۩Ĭő ʖĬ˃ɆĄɆĬʒĠ۩ĞĞőʋĬōɌɆĬʓȂʒĖʖĞōʳĞʒĕʓ۩ľőʖĄʒƂ˃ĠʡĬɆĄʯĄȿʬĄŔˌ Ą۩ĄĄʮő ĕĕʓēĖ ĄĄǶʳűĞé Ğŀ ʋĕ˃ɐʳĞ űĞĞŔċé ʋĞŎŀ ʡĄʒĄ˃ǶĄʳĮ ʖĕʢĕɐʓ۩ēʯĖ ʮĞő ŋĄ˄۩ʡĄɆĄő ʡĬɆĄő۩ʫĮ īő ʖĬ˃ɆĄɆĬʒĞɆĬʦ ĴɌɮȿʗĕő Ĵ˃ɐōʳĞɮʒʳĮ ʖĄɆĬ˃Ġ۩ĞĠ ăć۩ĞĞőȂʒĕő۩ŀʡĵŭèʑĬōʬĬő۩ĮʸĕȁʖĬ˃ɆĄɆĬʒĠ۩ĞĞőʧĬ˃ĬŎĄʬ ˄Ąʌ˄ Ľè ʖʣĬ˺ͼĬʒʍ۩ľő ʸĽřè ʫĄȿʌĄ ʡĕŘȋʍ۩ľőīĞĞʗʣňőʫĄʓèʋĄ˃Ğʺ۩Ĉ̐LJʣʹňˇʗ́ʺ۩ĈʡĄʍʲĕ˃ɎűȿʗĠ űʗćƁő۩Į ăʒĞŰʲĽĞʓ۩ĠƍȟʡĄʓ۩Ĉ۩ĆʯĄćŋĠʖĄĞĞőƁʬĬʳĮèʡĕɮő۩ĞĞĕő ɏʗĕő Ȃʒĕȟ ʡĬʬĬ˃ǶĄĠ Ĵɮʒĕʓ۩ľő ʡĬǶĄĠ۩ĞĞő ʡĄʢĄǶĮ۩ʫĄɆĄő ĝʗĠ Ĵɮʒĕʓ۩Ġ ʋĄʗĠʡĄɆĄʢĄĠűĞĞȿʗĄőéʡĕ˃Ė ʋĞŎĮȾĬĬʓīʬĬ˃ĞʬĬȿʯĠ
ĕʖ˃ĄĬĠĞő ĬĄ۩Ğļ۩۩Ğʡŭ ăĠ ʡĵŭ ļʺ۩ė۩ĞĞő ʡĬǶĄĠ۩ĞĞő ʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ
ĝʒĕ˃ȃĞĞʒĕȟ۩ĞĠʜĄɆĄ˃ʬĄʬĄȽůƁʗĬʦȂʒĖʖĞōʳĞʒĕʓ۩ľőŋĬʍ
ʸĕȁʖĬ˃ɆĄɆĬʒĠ۩ĞĞőʸĄǜőʖĬʬĬʒĬʓ۩ĬőʡĕŘȋʍůʲĕɮő
īĞĞʗʣňő۩ʡĕĠǵĄʌʢĮăʳĞʔĄǶĮ۩ʫĮŋĄʬĄʬĮ
ĕʖ˃ĄĬĠĞő ʒʖĞōʳĞʒĕʓ۩ľő ȂʒĖ ʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ ŵ˄èīʬĬ˃ĞʬĬȿʗĄőīʹĬʢĄő۩Ćʯć۩űő
ĕʖ˃ĄĬĠĞő ȂʒĖ ʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ ċé
Į ʲĵŭ ʓ۩Ĭő
ů
xv ĄĄ ۩Ġ ȟ Ą۩۩ʯŋăĄĬȂĕʯȟĞʮĞǶĬʒĄʯĞʒ Ȃʒĕʒʯĕȟ۩ĞĠ 㪥˃Ĭő ŋĠ ʡĄʓ۩Ĉ۩Ćʯć ʖĄŎĄɃċ۩ĞĠ ƍȟ ʋĵőʖĄŎĄʓʖĞɮʌʳĞʒĕʓ۩ĞĞĕőȂ˃ĕɌʢĕĠļʺ۩ė۩ʫĮʮĄȽʡĬǶĄĠƍ˃Ġ ĴɌɮʬĕɊűĞĞȿʗĄőéʧĕʪ ĄĞʋʬ˃ʌĠ ĞʒĄʢ۩ĄȠľʡʒʋŎĴɮʒĕʓ۩ľʦ ʲĬȋʹĄőűĞĞɆĬʒĄʒʢćèȂ˃ĕɌʳĞɮʒʳŀƁʒǶĮă˃ɃċʲĄʍ۩ĬőʡĄʓ۩Ĉ ʋĞŎŀ ʡĬʒć Ƞʌľő īĞĞʒɆĄʒʢć۩ʡĄĠ ʋĕʬĕ˃ĕʌʳĠ ŋĄ˃Ğő ʢʯè ĬʹʌĠĞʖĞʳ˃ɆʒĮʫʖĞőĞĄŋĵʒľɐʒĕɌ ʖĄĞĞő۩ĞĞĄ˄ ʖĄĞĞʳĞ˃ĄɆĬʒǶĮ۩ʫĮ ăʳĞʓ ʫĬ˺ʣĬ˄۩Ĭő ʲĬȋʹĄʌʳĠ۩ĞĠ ʖĄĞĞő ƍʢľʯĖè ŋĄʌĄĞĠ ʋĄŎĠ īʬĬ˃ĞʬĬȿʯĠ ǵĬʹĄ˄ ʸĬʍé Ş˃ɆʒȿĄ۩ʧɆɆŷ Ĭʬʬēʢőŀ ĬɌĬȂʒĕő۩ŀ ʋĬōɌɆĬʓ ē˃ʢĕő۩ŀ ʧĬʌʬĄʬĮ ŷ ʧĄɆĄɆĮ 㪥˃ĬɆĬʒĬȿʗĄő۩Į ēŎĖ ŹʯĞȟ۩ĞĠ ʋĄʹĄȽ۩ īĞĞʒɆĄʌʳĠ۩ʡĄĠ ǵĄ˃ĞʯĄōɌɆĬĠ ī˃ĞʬĄȉ۩Ćʯć۩űő ʲĬȋʗʢĄĠè Ŏŋʮī۩ʯ۩ő Ğŀ ĬĞɃċ űĞĞȿʗĄő ʲĬ˃ĞʒɃċ ʋĞŎŀ īȽ۩Ćʯć۩űő ʮĮ ŋĠ ʯȟĞ ĄňĞĄ۩Į ɌĕʖʬʒĄ۩ʧɐ۩Ⱦ˃Ąʗő Ȃʬľő ȾĄ˃ɆĄȿʗĄő ʧĕɐ˄۩ė ʖĬʬĬʒɆĄő۩Į ȊɌʗĕő ʋĄʣň˃ĞʹĄʓ۩ʫĮ űʗć ȾĄʬĄŔċè ʲĄʗĄǶĮ۩Ćʯć ăʒʬĄɃċ۩ŋĬɆĬʦ۩ĞĞĄő ɮĕ۩ʦľĝĕŀ ĞɆʌĠʡĠ ʒĮʫŋʒʹʦ ŋĵʒľʒĕ ŋĕʒĞʹĕʦ ƁʒǶĮ۩ʫĮ īĞʒɆĄʌʳĠ۩ʡĄĠ ĝʒĕɮŀ Ĵɮʒĕʓ۩ľʦ۩ľő ĄĄʬɆʦ ĬʗǶļĠ ĵʡǶĠĞő ʬʓĞĄűʗć ƁʬĬʓ۩ĞĞĄő ʡĬǶĄĠ۩ĞĞő ʋĵő ļɐĠ ʲĬȿʗĄǶĮ ʜĄɆĄ˃ʬĄɆĄʦ ʳļĕŀ ʒŔċēőēĖ ĄĄʖ˃ʠ űȿʗĠ۩ĞĞő ʖʣ˃Ğʠ ȾĄʬĄŔċ ēɮő۩ēʯĖ ƁʒĬŔċé ļʳĕʳŀ ɏʳŀ ʒ˃Ğʒȟʫʖ˃Ąʗőŷ Ĭʖ˃ʠ űȿʗĠ۩ʫĮĴŶ˄۩ľő ʋĞŎŀ ʖʣ˃Ğʠ ʲĬő ʡĕ˃Ė ŷ ʖĬ˃ɆĄȿʗĄőé ʡĬʒć ĝʒĕ˃ȃĞĞʒĕȟ۩ʫĮ ʮĞ˃ʌĄĞĞȿʗĄő ʸĕȁ īĞĞʒɆĄɆĬʒǶĮ۩ű˄ ƁʬĬʓ۩ĞĞĄő
īʬĞ˃ʬĄȿʯĠűȿʗĠƁʒǶĮʋĄʝćƍʢľʯĖè
ĄĞŋĞɐő ʲĄ˃Ğʍ ű˄ʋĄʗĠ۩ ĄʬĄȽè Ɍʗĕő ĞĞĄ˄ ľő۩ ʗĕő ʫĸ۩
ăʒʬĄɃċ۩ĞĠʲĄʗĄʳĮèűʗćʋĕ˃ɐʳĞʓ۩ľő ʲĄʬĄʒĄőʖĞōʳĞʒĕʒʢĖ۩ĞĠʋĞŎĠʡĕɮő۩ŀʖĄŎĄʓƁʬĬʓ ĄĄĠ ĝʒĕɮŀ ŋĄʌĄĞĠ ǵĬŸ۩ʫĮʖĄĞĞőīĞĞʒɆĄʌʳĠ۩ʡĄĠʡĄĞĞʒƂ˃ĞʒĄʳĮèăŘʁĠ ĴɌɮɌʗĕőéŋĞɎīʬĄɃċ۩ĞĞőīǶǴ۩Ĉ۩ʡĄĠ ʮĞʬĄȿʗĄőéǵĬ˃ʯĞő۩ŋĬʢĬȽ۩ĬőăʹĄʒɆĬő۩ĮʖĬʬĬʒɆĄő۩ŀăʳĞʓ۩ʫĮ ɐ۩ő ĄƁĬēʡňʗ۩ő ʨ۩۩Ğŋĵʒľɐő۩ŀ īʨǴ۩Ĉ۩ĞĞő ʡň˺ʗʠ۩ľő ēȁ ƁʒĬő ʖĄř ļɐʘ۩ľő űĞĞʯĄɆĄʒʢć۩ĞĠăʌʳĞʒʢć۩ʡĄĠɭĞʯĕʨɐʳŀʮĞʬĄǶĮ۩ű˄ļɐĠűĞĞȿʗĄő ĄĄĮ ĕʡʗƁĬʗőʡĕ˃Ė ƁʒĬȿʗĄőé ʡĬʗć ʸĕȁ ŋĄʬĄʬĮ ĕĖ Ĭʬ۩ʯŷ ʧĬʌʬć۩Ćʯć ʡĕ˃Ė ōĄĄʧ˃ʨĄűĞé ĕĖ ĬĄ۩ő ĬĬʖĕ˃ĕʌʳĞ ʖĬŎĬ˄ ǵĬʹĄ˄۩Ĭő űĞĞŔċé ʡĕʬĖ ʧĬ˃ĄʨʌĄ˄ īōʯĄʒĄő īʒĬȿʗĄő ʮĄȽ۩ĞĞĄő īǶĄɆĄʒĄȽè ŋĄʌĄĞĠļƎ˄ʋĬōɌɆĬʓ۩Ĭőļōʬľʗľʢĕő۩ŀʸĕȁʖĬ˃ɆĄɆĬʒĠèʡĵŭ۩ľ ʒĬ۩ʋʣ˃ʹʓĠ ĄĄĄʖōĞĕ۩Į ĬɆ˄Ĭɭō ʧĬʒɆĄ˄۩Ĭő ʖĞōʳĞʒĕȟ۩ʫĮ ʲĄʬĄʒĄő ʋĄʣň˃ĞʹĄʓ۩Ġ ăʹĄʒɆĬő۩Įʹ ōľľĕ۩ʸȁ ĬɆɆʒè ĵĴ˄ ĬɌĬ۩ő ʸĕȁ ʋĬōɌɆĬʓ۩Ĭő ĴƎ˄ ʋĵő ʖĬ˃ɆĄɆĬʒĠè ʸĕȁ ļōʬľʗľʢĕő۩ŀ ʖĬ˃ɆĄɆĬʒĠ۩ĞĞőŋĕʒĞʹĕʦīʒĄőȠʌľʘʡĬʘʡĬʘ ĄĄ ۩Ġ ĝʒĕɮŀʖĄŎĄɃċ۩ĞĠ
ʒĄʳʢĠ ĄĞĄ˃ʒɆʒĮ ĄĄĄʖĞōʳĞʒĕɮʒʳŀ ʲĄʬĄʒĄő ʡĄĞĞʒĄƂ˃ĞʒĄɆĬʒʳĮ ƁʒɆĄʌʳĞʢĄĠ
ŎŬĞĄŹĞŀ ʮĞʬĄȿʗĄő ŹʯĞʳŀ īŎĬŬʯĞʢĄĠ
è ĬĮ ʡɮ۩ʜŸ ēŘɐŲʹĄ˄۩Ġ ʜĄŸ ʡĕɮő۩ŀ ů ʖĬʬĮ Ūè
ĬĄĄʳļĠŵŷ ǵĬʹĄʬĬɆĄ˄ ŷ ļɐĠ۩ŵ˄ ʡĬʗĄʒĄʌʳĠ
ʲĽɊ۩ēʯĖʲĄ˃ĞʍʡĬȋʹĄʒʬĬʓ۩ʫĮ
˄ʯɮĴɮʒĕƎ˃Ġ۩ĞĠ Ȃʯĕɮŀ ŵ ˄
ʧĄɆĄƂ˃Ġ۩ĞĞőۯñʬĄȽۮ
īʒĄʍǵĬ˃ʯćƁʒɆĄʳĮ ʧĕʌʳĞʳŀűĞĞȿʗĄőé
ī˃ʯĞɆĬʒɃċ۩ĞĞő ʸĬʍé
ǵĄʒǴ۩Ĉ ʯĕʯĕɊ ʢĕĠè ʋĄŎĠ ő
xvi űʗćȾĄʒĞȿʒĄǶĮ ĕĖ ʡĬʗĬʍ۩Ĉ ʡĕ˃Ė ĞʯɆʒćĞő ʺėŵȾĮ Ɇʢ˄ǵĬè ĞĬǵĬĮ۩Ĉè ʖĞʹĬĮ ǵĬĬʘè űɆĄʢĬ˄è ȾĬĮ ļʺ۩ė۩ŵ˄ ʧĄɆĄɆĬʒʳĮ űĞĞʯĄɆĄʒʢć۩ĞĞő ŋĄʌĄĞĠ Ƞʌľʘ ȾĄʳĄ ʖĄ˃ĄŔċ ʲĕ˃Ɏ ȾĬˇĄőè ʡĬǶĄĠ۩ĞĞő ʸĬő ȒȃĄɆĬʒćè ʡĬʗĄʒĄʳĮ ʮĬȿʒĄɆĬʒǶĮ۩ʫĮ ʋĄʣň˃ĞʹĄʓ ʸĄƂʳĮ Ƞʌľő۩ʡĕĠǵĬʒƂʳĮĴɌɮʳŀűĞĞȿʗĄőéʮĬȿʒĄɆĬʒĬȿʗĄő ǵĄ˃ĄʍļɐĠʖĄŎĄʓƁʒɆĄȿʗĄőʡĬʗĄʒĄʌʳĠ۩ĞĠȠʒĞʹĕǶʳŀ ʑćīĞĞʒĄɆĄȿʗĄő ʳʓʫȂŀ ʬő ʌʘ Ĭŋĕʌĕ˄۩ĞĞĕő ǵĬŸ Ƞʌľʘ Ȃʬľő Ȃʬŀ ăʳĞʓ۩ʫĮ űĞĞǴċéȾĬ˃ŲʬĬɆĄ˄űĞĞʯĄɆĄʒʢćēȋʒĕȟļʺ۩ė۩ʫĸʮĮŷʖĬʬĮ۩ĞĞő ʸĕȃŘɖȾĄʒĞȽ۩ĠǵĬʹĄ˄ȠʌľőȊɌʗĕőéǵĬ˃ʯĞőăʹĄʒɆĬő۩ĮŋĕʒĞʹĕ ĄĞʖĞǶő źĄƂĠ Ľ۩ĞȾʒȽĞĄȾĄʒĞȿʒĄȿʗĄő ȾĄʒĞȽ۩ĞĞĄ˄ ʲĽĠ۩ĞĞő ăźʹĄʒƂ˃Ġ ʖĄĞĞǶĄő ŋĄ˃Ğő ȂʗĕɊ ǵĬ˃ʯĞőʋĵ˃ʢĕɐőʧĬʌʬĬʬĮǵĬʗĞɆĬő۩ĮȠʌľőŹʗĠƁʒŲʯĮʡĬř ĄǴĈĆĄɆőĮ ĕʋʣ˃ʹʓĠ ȿĄĄĄ۩ʲĽĞʯĠ۩ĞĞő īȿʗĄɆĄʢĄő۩Į ʋĄʣň˃ĞʹĄʓ۩Ġ ʸĕȁ ǵĄʒǴ۩Ĉ۩ĆʹĄʒɆĬő۩Į ˄ʋĕ˃ɐʳĞʒʢĕĠʋĞŎŀʖĄĞĞőŋĄĞĞʳćŪèʲĬʒʳĞ˃ɆĄʓű˃ĄȽȾĄȿʯć۩ű ĄĞʒȿĄé Ĭ˃īĞʣĞ۩ő ăĬʣĬʌĄʢʯĞʒĄő īĞĞʗʣʒĞř۩Ĭő ˂ĬƖ˃ʠ ȾĄʒĞȿʒĄȿʗĄőé
ĄňĞĄ۩ăʺĈ ʲĽĞʯĠ ăźʺ۩Ĉ ʋĄʣň˃ĞʹĄʓ۩Ġ
ŋĠʡĬřʮĞǶĬʒćăʳĞʓűĞĞȿʗĄőŋĠʋĕʬĕɐʳĠé۩ēŎĖ ĬʲźĬļĠ Ğĕɐ۩˄ űĞĞȿʗĄőé ȠĞʯĕʨɐȟ۩ű˄ ļɐĠ ʲĬźɆĬ˄ ǵĬĮ űĞĞȿʗĄő
ƁʓăʳĞʓ۩ĞĞĄőīʒĄʍăʌʳĞʒʢć۩ʡĄĠƅɠʦ
ĄȾĮ ĬŋĴŶ˄۩ĞĞōő ŋĠ ʲĬő ȾĬĮ ǵĄő
˃ɆʒȿĄʮĮè ī˃ĬɆĬʒĬȿʗĄő
ăŬǶĮȂ˃ĕɌʢĕĠ۩ĞĠ
ʋĄʣň˃ĞʹĄʓ۩Ġ ʡĕʌʬĕɌʒĕʳŀ ĞĞĄő۩ Ĭő۩˄ é ʦ
˃ʗĬ˄ ʣĬè Ɓ˃ʳĞɘő ăʣˇĬʦè ˹˃ĞʗʣĬ˳˄ ʹōɆʢ˄ˠʒ̣˄ ƁʹĄōŲɆĄʢĬ˄è Ĭʢʌʹ˄ʊűĄʗĞǶőʭȾōĬɃċŇć ʮĄʣĩè Ňňć ȾĄōʢĬʒɃċè ʭè űʹĄ˃ʗĄĞĞǶĄőè ʊè ʡĬȿʢĄʌŲʹĄ˄è ĞʳʒʓľăĄĬĄ۩ő Ą۩۩˄ ĵĕʳĴĶɌɮɌʗĕőé ʲĵƘĕʌʳĠ ʡĄʓ۩Ĉ۩ű˄ ăʗĄɆĬʬĄʓ۩Ĭő ʖĞōʳĞʒĕʓ۩ľő ˃ʨĞĄʖɮʳʒɌĠ ʒȿĄͻōĞƕ˺ĞʠŋʬʬȂʒ ƕ˃˺ˇĞʗʠ ŋĄʬĄʬĮ ͻňō˃Ğ˻ ƁʒĬȿʗĄő ʖĞɮʌʳĞʒĕɌʯĠ ī˃ĞʨʯĞʒĄő ĵĕʳů ĬʲȋĄļɮŀ ĞȿĄ۩Į ĕűʹĄ˃ʒĄʳĮ ʸĕȁ űĞĞȿʗĄő۩ʫĮ ļɌɮʳŀ ʲĬȋʹĄʓ ŋĬʍ ů ĴŶ ʲĵƘĕʌʳĠ ʋĞŎĮ ƁʒǶĮ ʖĞōʳĞʒĕʒʢĕő ȂʒĖ ʡĬʗĄʒĄʌʳĠ ʡĕ˃Ė ʸĄƂʬĄȽé ĬĬɆʢ˄īĬĄʳŭȒʳʍľʖʬʒʓĬȾĮĈ ů ȾĬĮ۩Ĉ ʖĬʬĬʒĬʓ۩Ĭő ȒɌʳĞʍ۩ľő ī˃ĬʒĄʌʳĄŭè ʖĬʬĬŲɆĄʢĬ˄è ŷȂʬľőȠʌľʘ۩ĠʧĬ˃ʬĄʺ۩Ĉéƍ˃ĞōʗĄĞĞőèī˃ĄōʯĞʌĕɊèīʹĬōɐ˃ĕʓ ʡĄʓ۩Ĉ۩ű˄ ʡĕľȁ۩ĞĞő ů ŋĕĞĞɐʍ ǵĬ˃ʯĞő ŋĄʬĄʬĮ ʧĵ˃Ŵ ʡĄĞĞƂř Ȃ˃ĕɌʒĕȟ۩ŵ˄űʢĮʖĞĞʬľɌʗĕőűĞĞɃċé ʳŷ ɐʗʳʮʬǶļɐĠé ʮĞʬĄǶĮ ļɐĞʗȃʳŀ ŷ ɏʳŀ
ƍʢľɐʓ۩ʫĸƍ˃ĞőʲĬȋʗʢĄĠȂʌʳĞʺ۩ė۩ŵ˄ă˃ĄĠī˃ĬʳĮʮĞʬĄɃċļɐĠ ˄ƁʒĬŬʗĬ˃ĄʒĬȿʗĄőǵĄʒǴ۩ĈʋĄʣň˃ĞʹĄʓ۩Ĭʦ۩ĠŷʫĬ˺ʣĬ˄۩ĬőăʳĞʓ۩ ʡĕʬĕɐ Ȃ˃ĕɌʒĕʳŀăʌʳĞɃċļɐĠ۩ʫĄɐőʋĄʗĠʸĕȁǵĄ˃ĄʌʗĄʳĮűĞĞŔċè ĄĄĠ Ŏʡ˃ĬʹʒĬ۩Į ĕʳɌĕƁĬē˃ʬĕʌʢĕő ƁʗĬʦ ʧĕʌʳĞɌʗĕő ʡĕ˃ĖʡĬȋʹĄʒʬĬʓ۩ʫĮ ēŎĖ ŋĄʌĄĞĠ
ĄňĞĄ۩ŷ ˃ɐľ۩Į Ȋȟ ĝ˃ĕɐʬľĠ۩ʫĮ ŷ ʋĄʣň˃ĞʹĄʓ۩Ġ
Ğʯè ū˃Ğʣć˂ĞŲʯ̇è
ĄĄè ŬĬĞĞʸĕ̑ĞLjè ăŬʬĬ˃ĞʝĞʦ ʋĄŤĄĠè
˄Ɩů ƍɮʦ۩Ćʯć ů Ɍ˃ĽƖ˃
ĬĬɆőĮ ăʳĞʓ۩ʫĄɆĄőʖĬʬĬʒɆĄő۩Į ĬĄè űʹĄōʯĬȿʢĮè ʧĬ˃Ąőè ǵĄ˃Ğő ʫĄɆĄő
ēʯĖ۩ ľő۩˄ ʋĞŎŀ ʐè è Ė
xvii ĞȿĄé ĕĕ˄ľʧʌĞʗő ĬĬɃċ ˃ɐľ۩ĸ ƍʢľŔ ʖĬʬĬʒɃċ ĝ˃ĕɐʬľĠ۩ʫĸ ʧĕʌʳĞɌʗĕő ʡĕʬĕɐ˄۩ľő űĞĞȿʗĄőé ē˃ʬĕŎĞʌĶŘɎèȠȁè ʳʓĬļ۩۩ĸ Ğŀ ʗʦ ĬĬɆőĮ ĬĞǶʓĠ ʧĕʌʳĞʳŀ ʖĬŎĞ˃ǶĄʓ۩Ġ ʖĬʬĬʒɆĄő۩Į ƁʗĬʦ ʋĞŎŀ ļʺ۩ė۩ʫĸ ăʳĞʓ۩Ĭő ʡ ʖĬʬĬʒɃċ۩ĞĠ ƁʒĬʌʳĠ ēŎĖ űĞĞŔċé ʋĞŎŀ ŋĄĞĞʬĄʳĮ ɏʬĕɊ۩ʡĸ ŋĬʍ۩ĬőʖĄř۩ĬőȾĬĮ۩Ĉè̀˃ʗʣňőˋ˃ĞĞʬʼnèȾĄʒʗĄřèʡĬʒɆĄɆĬ˃Ġè ˄ĝʒĕɮŀ۩ʡĕĠ۩ŵ ĬĄĄĞĄȠĕŲĄĬè ĄʯʌɊ ĬĄăĄĄè ʲĬȋʹĄő ăʹĄʒĄʓè ǵĬʹĄ˄ ǵĄʗʯĞʌĕɊ Ƞ˃ĕʒŲɆĄʢĬ˄è ʡĬʗĄɆĄĞĞʒĄő ŋĄʬĄʬĮ ʧĕʌʳĞɌʗĕőé ŋĄʌĄĞĠ ăʗĄɆĬʬĄʓ۩ʫĮ űĞĞɆĬʒĄʒʢć۩ĞĞő ʓ Ŏʫ˺Ĭ۩ő ʳʓ ʡĕʘ ăʳĞʓ ʫĬ˺ʣĬ˄۩Ĭő ēŎĖ Ɓʓ ʖĞĞʬľɌʗĕőůʋĵő ĕĕʯȿĮ ȿĠŋ˄ʡɆő ēñōʬĖ ŷ īő۩ĆʯćēȋʒĕőʧĕʌʳĞʳŀèʲĵƘĕʌʳĠĴɌɮʳŀʸĄƂȿʗĄőʖĕʯĕōŲɆĄʢĬ űȿʗĠ۩ŋĄ˄۩ʡĄɆĄő ʖĕʯĕōʯĬȿʢĮ ĕʡʗʒʳűĞʗőļ˄ ĬɌĬ۩ő ĕʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ ʋĬōɌɆĬʓ۩Ĭő ļƎ˄ űĞĞȿʗĄőé ʡĬʗĄʒĄʳĮ ʸĕȁ ĬĄǶļɐĠ ʧĬ˃ĄʨǶĮ ĵľȉ ʒʢőʋĵő Ȃʒĕʢĕőè ʡĵ˃ľʒȉ ʡĄʒĄ˃ǶĄʺ۩Ĉé
ēñōʬĖʧĬ˃ĄʬĬɃċ
ĝʒʹćɌ˃ĽōʣĬˈŋĠ۩ŋĬʍĴɮʒĕʓ۩ľőʡĄʓ۩Ĉ۩ű˄ŋĄʬĄʬĮ
ĞɃċ ʗŋɎ ʒɊ Ƞʌľʘ ƍʒľɊ ŋĞɎ űĞĞɃċ űʗć
ļɐĠè
ʧĬ˃ĄʨǶĮ۩űő ʱèűʹĄōʗĄřè
ñʬīʯʒő ĬĄǶļɐĠ۩ŵ˄ ʧĬ˃ĄʨǶĮ īōʯĄʒĄő ēñōʬĖ Ȃʒĕő۩ĮŵőʡĬǶĄĠʋĕʬĕɐʒĕʓļɌɮɌʗĕő
ōĬƁǶűĞʗőēñōʬĖ űĞĞȿʗĄőé ƁʒǶĮ īōʬĬĮ ʋĄ˃ʢĄȿʗĄőȠʌľʘűĞĞɃċ
ȠȃŲɆĄʢĬ˄èʋĵőĝʒĄŢĬʺ۩Ĉ
ʗǵĄʒĄɆĬő۩ĞʽĄ˄ űʗć
ʫĬ˺ʣĬ˄۩Ĭő 2006
ļɐĠ ˄è ĄĠ ę
ʒőij ĬĬĬ۩Į ŘʯȃīĬʯȿĄʋŎŋʬʬʮĞǶĬʒć ŋĄʬĄʬĮ ʋĄŎĠ ī˃ĬʒʯĄȿʗĄő ĴŘɐʯĕȃő ʖĬʬĬʒĬʓ۩ʫĮ Ȃʒĕő۩ij ʡĬ˃ʗĞʒɃċƁʒĬȿʗĄőƅɐʦēŎĖʖĬʬĬʒĬʓŋĠǵĬĞĞʗĞʬćʸĕ˃ĶŘȃĠ ʖĞōʳĞʒĕʓ۩ʡĕĠĝʒĕɮŀʖĄĞĞőŋĕĞĞʒĕŶʓʲĬȋʗʢĄĠůĝŘɐȿʗĕő۩Ğ ʖĄʒƂ˃Ġ۩ʡĄĠ ɐ۩۩ĞʖʬʒĄ۩ƁĬȠľăĞĄ۩ʡŎő ʋĕʬĕȡĠ۩Ğ ʡĄŎĞő ăʌĞʢĄő۩Į Ƞʌľő ƁʒĬő ʖĬʬĬʒɆĄő۩Į Ȃɐ˄۩ė۩ĞĞő ʖĬʬĬʒĬʓ۩ʫĮʸĕȁŋĕʌĕ˄ƁʒʳĮʮĞʬĄǶĮʖĄʒƂ˃ĠʸĬʍè ŋĄʬĄʢĄĠ Ɓʓ Ƞʌľʘ ƍ˃ĞʬľɮʒľɌʗĕő ŋĵȡʯĕʓ۩Ġ ʡĄɆĄʢĄĠ ʲĽɊ۩Ćʯć ŋĄĞĞʬĄʳĮűĞĞŔċé ĞʯɆȿĠ̐Ǫň۩ő ʒʖōĞĕʢőʧĬȋʹĄő۩Į ʖĞōʳĞʒĕʒʢĕőè ȂʒĖ ̐ʌǪʢň˄۩ľő űĞĞʯĄɆĄȿʯĠè ʖĬʬĬʒɆĄő۩ĮȾĄʳĄ˄űĞĞʓ۩Ĉé̐ʌǪʣň˄۩ľőȂʒĖʖĞōʳĞʒĕʓŋĠƁʗ ƍɮő۩ŀȊʯĞʹĕő ʡĕŘȋʍ۩ʫĸ ĬĬĬʯè Ą۩ĞʮȽ ʲʌʘ ȡȂĖ ʖĞōʳĞʒĕʓ۩ľő ȂʒĖ ƍȡĠ ʲĞʌĞʘ ů ʮĄȽ ʮĄĠ۩ĞĞő ʖĬʬĬʒĬȿʯĠè űĞĞŔċéēŎĖʖĬʬĬʒɆĄő۩Įļ˄۩ėʫĽřè ʒĄʗő ȟ ʌʘʡŷ ĞʖʨʓēĖŵʡĄʒĄ˃ǶĄʳĮ ʖĕʨȋʓ۩ēʯĖ۩ŵő ʮĞő ŷ Ƞʌľʘ۩ʡĸ ƍȟ ƁʒɆĄȿʗĄő ʣ˺ĬʍĬʸȁ ʖĬ˃ɆĄɆĬʒĠ ʸĕȁ ʖʣĬ˺ͼĬʒʍ۩Ĭő Ĭʣ˄ĬăĞ۩Ğő ʯȟʫǵŸĞő ăʌĞʬĬ˃Ąʓ۩Ĭő ǵĬŸ۩ĞĞő ŹʯĞȟ۩ʫĮ ăʳĞʓ۩ĞĞĄő ʫĬ˺ʣĬ˄۩Ĭő
ăʳĞʔĄǶĮ۩ʫĮŋĄʬĄʬĮȊʒŲ˄ʡĄɆĄʢĄĠèīǶĄɆĄőʖĕ˃ɮɌʗĕő ĄĞǵʌć ăʳĞʔĄʳĮ ǵĄʌʢć ʖĄĞĞő
ʮĞ˃ʌĄĞĞʒʢć۩ĞĠʸĄʌĄ˄ŋĞɐőűĞĞʬĄʓ۩ĞĞĄ˄ƁʯĄɆĄŔċ
ŋĠ ĞȿĄé ʗʧĬȋʹĄő۩Į űʗć űĞĞȿʗĄőé
ŭ Ğʳʯè Ȃɐ˄۩ė۩ĞĞő ʖĞōʳĞʯĠè Ȃŭ
ăʳĠǵĬʒƂȿʬĄʓŋĠʡć
Ğĕ˄۩ĞĞĕő Ȃʒĕő۩ŀ
ȂʒĖ ɏʳŀ Ĭʦ Ğő
xviii űĞĞǶĮ۩ʫĮ ɏȟʋĕʢŀȠʯĞőʲĽĞʓ۩ļɐĠ۩ŵ˄ʖʣĬ˺ͼĬʒʍǵĬʢćļŎĕȃ˄ăʌĞʬĬ˃ĄǶĮ ʡĄɆĄʢĄĠȾĄʳĄ˄ƁʒʳĮʮĞʬĄǶĮ ʸĬʍʖĞȽȂʢŀʡĬĞĞʒʗĞ˃ĄȿʗĄőīǶĄʌʗĄ˄۩ʡĮǵĄ˃ĞɆĬʯĄʒɃċ۩ĞĠʡĬʒ ăȿƖʯĬĄʌňʬį ĵŇĞǭʬ˄˂Ɩʠ īĞĞʗʣʒĞřè ˂ĬƖ˃ʠ ǭ˃ʬň˄è ŇʌĞʓ ʋĵő ŋĵʒľɐʒĕʳŀűĞĞɃċŵ˃ʢĕɌʯĞőȒ˃ľŘɐʢĕőŋĄ˄èʜĬ˃Ĭř۩ʡĮűĞĞɃċ ĬĬĄĄ۩Į ĕʖ˃ĄĬĠĞő ʌʓĝĬè ŋĞ˺ʒĄʘ ĝˇĬőè ŇʌĞʓè ʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ ǵĬɆĬʯĄɆĄő۩ʫĮ ʋĄ˃ʹĄʣʣć īĞĞ˃ʗʣ˃ĬĞĞʍè ˄ŎȃīĄ۩Į ĄĞĄʖʒʳʮʬǶƁĬʗőĝˇĬő ƁʒĬȿʗĄő ʮĞʬĄǶĮ ʖĕʒĕʳŀ ʖĄĞĞǶĄő īʗĄő۩ʫĮ ļŎĕȃ ǵĄʌʢĮ ܅ʡĕɮʘ۩ė ĄȋĄĮ űĞĞȿʗĄőéʧĄŬȋʯĄʳĮ ʖĞʌǮ˄ ʖĄɆĬʯć۩ĞĞőăʗĄɆĬʬĄʓ۩ʫĮ ʒő ˄ʡǵʌĮ ĄĬĮ ăʳĞʔĄʬĄȽ ʖĄɆĬʳĮ ǵĄʌʢĮ ɏ˄۩ʡĸ ăʒŲő ĄĞĬĄɃċ Ġ ʒʍ ɆʗȿĄè ĬćĞő Ğ˃ī˃Ĭő űĞĞ˃Ġ ǵĬʢć۩ĞĞő űɆĄʗĬȿʗĄőè īʒĄʍ ŋĠ ǵĄ˃ĞɆĬʯĄʒɃċ
ʖĬ˃ĬʒʯĄȿʗĄőʫĬ˺ʣĬ˄۩ĬőīʹĬʢĄő۩Ĭʦƅɐʦʡĕʬĕɐ˄ŋĠʋĞŎŀ ʒőʫʗʒř ĄĬ ʧňʗŁ è ͻĄɆĬʘ ʫňʗĄʒňř ăʒňőè
ǵĬɆĬʯĄɆĄő۩ʫĮ ĝŸ ůˋ˃Ąọ̄ʗ˿ʫĠȾĄ˃ƀ ˋ˃ć ʌʬ˃ǶīʯőĮ Ĭ۩Į Į ʗő ʡĽ˄ ɏʗĕő ʮĮ ǵĬŸ۩ʫĮ ī˃ʯĞő۩Į ăʌĞʬĬ˃ĄǶĮ
Ĭ۩ő ĄĞćīĬ۩ő ŋĕĞĞɐʍ۩ľő īʒĬʘ۩Ĭő ʮĄĞĞʳćè ҡ˃ĬLj۩ľőˇ
˄īʒĬʘʡĵ˃ĸ۩ĞĞőă˃Ƀċ۩ű˄ʡĬřʋĄɆĮ۩ű
ʒʍĈēɐĬĄè ȂʗĞɌʳĞ˃ɆĄʓè ēŘɐʢĬʹĄɃċè ăʒĞʍ۩Ĉè Ğʗ˃ġɏĕȠʌľʘ۩Ĭő۩Ȉ۩ʡĕĠ ɏʗĕő ʖĞ˺ʗʣ˃Ķġʍ űĞĞȿʗĄőƁʒŰ۩IJéīʒĄɆĄōŲɆĄʢĬ˄۩ʡĮ
ĬĄͻĞ˃˃è Ǔʺ۩Ĉ ͻĬĞĞ˃Ɩ˃Ġè ʡĬʌĄʘ
ʒƁĞʣʓů ƁĞĞ˃ʣňʓè ̀ʒʼn īʬĮ ĝʒĄōɌɆĬĠ۩Ĉ ʒōɆĠĈ
ʸĬʍéűʗć ɆĄȿʗĄő
ĄʢʒɏʋʢŋĄĞĞʳć ʋĕʢŀ ȾĄōʢĬʒɃċ ɏȟ űĞĞȿʗĄőé ʲĬʒĮ ۩ʫĄɆĄő ʲĬʒʳĞ˃ɆĄʓ۩ ŋĞɎʮĞǶĬʒćʸĄɆĬʍ۩ĈʧĬʢĄȿʬĄǶĮ۩űĞĞȿʗĄőéʖĬʬĬʒɆĄő۩ĮăʳĞʓ ĞĬʋʹĮĞő ĄĄĬć ĞǶƁĬĠ ĬĬĄŋʬʬʋĄʗ ǵĬɆĬʯĄɃċ ŋĄʬĄʬĮ ƁɆĬŎĠ űĞĞǶĮ ǵĄʳĄɆĬʬć ʋĞʹĬĮ۩ĞĞő ʋĞʹĬĮ ĴŘɐ˃ĕɮʒʳŀűĞĞȿʗĄőéǵĄʌĬȽ۩ĆʯćĝʒĕɮŀŷīʒĄőʲĞʓ۩ĬőŋĄĞĞʳ ļŎĞʢĕĠéʋĵőʫĬ˺ʣĬ˄۩ĬőăʳĞʓ۩ĬőƍȟǵĬɆĬʯĄɃċ۩ʫĮ ŵĞĞʳĞřǵĬʢć۩ʡĄĠǵĄ˃ĞʯĄōɌɆĬĠűɃċǵĬʒƂɃċ۩ʡĄĠűĞĞȿʗőƁʒŲʯĮ ĞŋɎ ʬɃċ ɆʒĄĮ ĞĄʗő ĞĬʸè ʋĵŘɐɐ˃ĕʓ ʸʼnè ʑĞʹĬĮ ʮĞʬĄȿʗĄő īʬĄɃċ ăɆĬʒʳĄʳĮ ŋĞɎ ŋĞɎ ăʒŲő ʡĕ˃ɐʓ ʲĬő۩Ġ ʖĄĞĞǶĄő ʒő ĕ۩ŋʲʒĄĄ ĕǵĄʌʢ ʲĬʒʳĄʹĄɃċ ʡĕĠ ŋĠ ŋĕ˄۩ė ĄĞĄʲőĠ ĕɐ ŋĄĞĞʳćŋĄʬĄʬĮļŎĖ۩ʡĕĠéʖĄʹćʧĬ˃ĄʬĬȿʗĄőȠʌľʘ۩ēʯĖă˃ĄĠűɃ ĬƂ ۩ĄűĞʗőǵ˃ő ʬĠ ŀ ĄʢĴŘɐ˃ĕɮʒľɌʗĕő ǵĄʌʢĮ ʮŀ ȂʬľĠ ǵĄ˃Ğő ǵĬʒƂɃċ۩ʡĄĠ űĞĞȿʗĄőè ʬő ʌʘ ʓ ĬĄǵĮĈƁǵř Ġʑő Ĝè ʑĞő ʑĠè ǵĬř ƁĮ ǵĬĮ۩Ĉè ǵĬʹĄ˄ Ɓʓ Ƞʌľʘ Ȃʬľő űʢĬʗĄĞĞǶĄőèļŎĕōʯĞŲɆĄʢĬ˄è ĬĄè Ĭʖĕʯĕőè ʲĬĮ ʮĬˇĄőè ʡĬǶĄĠ۩ĞĞőʮĄȽèʋĽʯĠ۩ʫĸʡĬʬĬ˃ǶĄĠȂʌʳĞʺ۩ė۩ŵ˄ʮĞǶĬʒćűĞĞȿʗĄő ʑĄ˃Ğʗʗćèăƀˆň˃Ŏň˄èēʳĠǏʣ˃ćè 2011
˃ʒʌĄŋ˄ ʒŔċʋŎĴő ʧĕɮŀ Ĵȋő ƁʒĬŔċé ʋĞŎŀ ŋĄ˄ ī˃ĬʒĄʌʳĄŭ őʫŷ ʌʬ˃ʓĬ۩Ğő ǵĄʌĬȽ۩Ĭő ăʌĞʬĬ˃Ąʓ۩Ĭő۩ĞĞĄő ŷ īő۩ʫĮ ĄĮ ĞŔċʮʌ۩ɐ۩˄ ŋĄʬĄʬĮ űĞĞŔċé ʮĞʌć۩ļɐĠ۩ŵ˄ űʹĄ˃ʒĄʳĮ˃ʹ
ĄʋŎăĞĬĄʖĞǶő ƁʒɆĄʳĮ ʖĄĞĞǶĄő ăʌĞʬĬ˃Ąʓ ʋĄŎĠ ŋĄ˄
ʌʗ˃řʲōĄȡè ć ˩ ʋć ʲĄōʬĄōȡŀè ŹʌŲʗľ˃ľřè
ʋĞŎŀēŎĖăʳĞʓ۩ĞĞĄőʧĄɆĬʗȿĄǶĮ
ʸĕ˃ŀ۩ʫĕɐő
ʋĞŎĠ ć Į ċ Ġ
xix ī˃ĬʒʯĄǶĮʲĄ˃ĬʬĄʓ۩ĠʧĄɆĄʳĮĴɌɮɌʗĕőé ʳʺėʮƁƁǶļĠ ĞǶűĞʗő ˃ʓ ʗő ŋĞɎ ɏʗĕő ī˃ɘʓ űĞĞȿʗĄő űĞĞǶĮ ļɐĠ ƁʒǶĮ ƁĠ ȂʳĞʺ۩ė۩ʮĮ ļʺ۩ė۩ʫĮʧĽ˃ŶɮʒĖăʌĞʢĮűĞĞɃċĴƎɎēʌĕɎůʡĕĞĞʌŀī˃ʢĮ ʮĞʬĄɃċŋĄʗĬʒĄʳĮ 2014 ȠʌľőűĞĞɃċé ʖʣĬ˺ͼĬʒʍ۩Ĭő ēĞĞʌĕ˃ȡŀ űĞĞȿʗĄőé ăʳĞʒʢĄő ʧĄɆĄʢǶĄʓ۩ʡĄĠ īʹĬʢĄő ʫĬ˺ʣĬ˄۩Ĭő ŋĠ Ğʌ˃ŀ ĬĬɆőĮ ŋĽȡ ʖĬʬĬʒɆĄő۩Į ĄĄǶʓʡĠ ʳʒĄűĞʗő ˃ ļĠʋĵőă˃Ƀċ ļɐĠé ʋĕʬĕɐʳĠ۩ĞĞőƁʒŲʯĮɻ˃ʌĄőīʒĬʘ۩ʡĮʸĕȃŬʯĞʒĕőűĞĞʬĄȽļɐĠŋĠ ĕȋŋĠʡĕŘȋʍ ۩ʖĄ˄۩Ĉ۩ű˄ʋĬōɌɆĬʓīʒĬʘ۩ĬőʸĕȁʖĬ˃ɆĄɆĬʒĠ۩ʫĮʲĬʯĞőʖĬʬĬʒĬɆĄʯĠ ˄ŎȃĝĞĕʲĞéēőēĖ ĄĄȂĖ ʖĞŎĕʳĞʒĕʓ۩ľő ȂʒĖ ȾĄʬĄŔċ ēɮő۩ēʯĖ ʲĽĞʓé ĝʯĞʌĕ˄ ļŎĕȃ ʋĞŎŀ ăʳĞʔĄǶĮʲĄ˃ĬʬĄʓ۩Ġʖˇˇőů Ĭʣ˄ĬȊĞĕ۩ő ʢő ĬĬĄĄ۩ʡ˃Ġʫŷ ʡĬ˃ʗĠ۩ʫĮ ǵĬɆĬʯĄɆĄő۩Į ƍʢĕő ȊʯĞʹĕʓ۩Ĭő ʫĬ˺ʣĬ˄۩Ĭő 2012
īő۩ʫĮǾ˃ĬŎĞŘňőǵĬʢć۩ʫĮƁʒĬȿʗĄőǵĬ˃Ąʓ۩ʫĮăʹĄʒĄǶĮè 1 .
īő۩ʫĮȒȃǶĬʢćè ĕʖ˃ĄĬĠĞő ʢő ĄĞ۩Ąè ŋĕĞĞɐʍ۩ľő ʮĄʒĞř۩ʡĄĠè ƍʢĕő ʖĬ˃ɆĄɆĬʒĠ۩ĞĞő ʸĕȁ 5 5
ɐʋĞŋĄ˄۩Ġ ʋňĞ̇ ļɐĠ ɐ۩۩ĞűĞĞʯĄɆĄʒʢć۩ůȂɐ˄۩ė۩ĞĞő
Ŏʫ˺Ĭ۩ő ʳʓ ʸĄʌĄ˄ ăʳĞʓ ʫĬ˺ʣĬ˄۩Ĭő ēŎĖ
ʌőʫʲĬȋʗʢĄĠ Ƞʌľő۩ʫĮ
ʫĄȿʌĄ˄ʖĄʒňōʘ۩ĬőʖĄřʧĄɆĄȿʗĄőů 2014
ĬĄǶ۩ɐűĞĞǶĮ۩ĞĞő ʧĬ˃ĄʨǶĮ۩ļɐĠ 2013
īő۩ʫĮȾĬĞĞʣĞŘňőů 5
Ľʓ ʬɊ ŋĠ ɏʬĕɊ ʲĽĞʓ
ĞǵĬ˃Ąʓ۩ʫĮ ī˃ʯĞʍ˃ īő۩ʫĮʧĬʒĬɆĄő
Į ȂʒŶ˃Ġ۩ŵ˄ʮĮ ʯĕʓ ű˄
ī˃ĬʒʯĄǶĮʲĄ˃ĬʬĄʓ۩Ġ˹ŎĽʠůăʒĞʘˆĄʒňōƖ˃Ġ۩ĞĞőʖĄřè̈Ąō Ąů ĬĮĞő Ąŋʡʘ ĬʧĄɆĄȿʗĄőé ʡĬʘ ʡĬʘ ŋĠ ʖĄř ȾĬʒĮ۩ĞĞő ů ʖĄř ʬň˄۩Ĭő
xx Introduction
The present thesis consists of this introduction - which among other things contains information on terminology, methodology, summaries and a histor- ical and areal overview - and the following papers:
x The aspect-evidentiality system of Middle Mongol. Ural-Altaic Studies, 13 . (forthcoming) x The tense-aspect system of Khorchin Mongolian. In: Pirkko Suihkonen & Lindsay Whaley (eds.), Typology of Languages of Europe and Northern and Central Asia . Amsterdam: John Benjamins. (forthcoming) x Aspect and epistemic notions in the present tense system of Khalkha Mongoli- an. Acta Linguistica Petropolitana . (forthcoming) x Factual vs. evidential? - The past tense forms of spoken Khalkha Mongolian. In: Ad Foolen, Helen de Hoop, & Gijs Mulder (eds.), Empirical Approaches to Evidentiality . Amsterdam: John Benjamins. (under review)
The purpose of the thesis is to give an account of Tense, Aspect and Eviden- tiality (TAE) in three Mongolian varieties: Middle Mongol (MM) as it was spoken in the Mongol Empire in the 13 th and 14 th century; Khalkha Mongo- lian as spoken in the Mongolian state (i.e., what is commonly called “Mon- golian”), which was influenced by Central Asian Turkic and Tibetan; and Khorchin Mongolian as spoken in eastern Inner Mongolia, China, which was influenced by Manchu and Mandarin. In a nutshell, MM started out with a simple tripartite tense distinction and a medium-sized aspectual system that showed traces of substantial recent changes. Its past evidential system was tripartite as well: it had suffixes for firsthand, non-firsthand and evidentially neutral information. In Khorchin, evidentiality was lost, and tense was simplified into a past / non-past distinc- tion, alongside with a discontinuous proximal future/past marker. The aspect system underwent some structural changes, but retained a complexity com- parable to MM. Khalkha, on the other hand, underwent some shared innova- tions with Khorchin, but retained participles as a multifunctional unit within finite predicates, thereby obtaining a rather complex aspectual system. The past evidentiality distinctions of MM were retained with slight modifica- tions, but the introduction of present tense evidentiality brought a number of changes: the evidentially neutral value shifted to signaling assimilated knowledge, and discontinuous future uses were introduced for all past mark- ers.
1 All papers were structured uniformly with respect to font, page size, text styles etc., although slight exceptions (such as the formatting of example sentences in the paper on Khorchin) have been retained. The texts them- selves have been left untouched, which means that differences in biblio- graphical conventions and to some degree transcription systems exist. The remaining part of the introduction is structured as follows: section 1 gives a brief overview of the classification of Mongolic. Section 2 gives a short, fairly non-technical introduction to the terminology used throughout this thesis, addressing both TAE and a number of structural terms that apply to Mongolian. In section 3, previous research on TAE in Mongolian is dis- cussed from a methodological vantage point. Before this background, the methodological approach adapted in this thesis is then introduced and moti- vated. Section 4 provides short summaries of the four studies that constitute this thesis. Section 5 gives an overview of how several elements of the Mon- golic TAE systems developed from MM to Khorchin, Khalkha and, as far as can be told from the literature, to the western Central Mongolic dialect Oirat. The sections 6 and 7 take a look at the areal context within which these changes occurred in Khorchin and Khalkha, respectively. Section 8 con- cludes the introduction with some conclusions about the TAE system of Khalkha as a whole, the areal context of TAE in Central Asia and the classi- fication of Central Mongolic. 1. Classification of Mongolic
In this study, I focus on Middle Mongol, Khalkha and Khorchin, so it is nec- essary to relate these varieties genetically. The most ancient attested Mon- golic languages are Tabghach as spoken in the 4 th to 6 th century CE and pre- served in 14 words found in Chinese texts (Vovin 2007); Khitan as written from the early 10 th century and preserved in texts from the 11 th to 14 th centu- ry (Janhunen 2012); and Middle Mongol as preserved in texts from the 13 th and 14 th century. While the taxonomic status of Tabghach within Mongolic has not been dealt with, Khitan is a sister to MM with no surviving modern varieties. Due to problems in deciphering the two Khitan script systems, only part of its morphology has been identified, and while we can identify the three past suffixes -ar ,-lún and -bún (with variants, Kane 2009: 144-7), little can be said about their precise semantic properties. For this reason, the term Proto-Mongolic (PM) will be used for the language that can be reconstructed from all modern Mongolic languages (Janhunen 2003) and MM, not for the mother of Pre-MM, Khitan and - in theory - Tabghach. While there is no variety actually called “Old Mongolian” (see de Rachewiltz 1999), MM itself can roughly be divided into Early MM, Late Western MM and Late Eastern MM. From the 15 th up to the end of the 16 th century, sources are
2 almost absent. Sources from the late 16 th to the 19 th century in Mongolian script are here referred to under the cover term Classical Mongolian .3 The modern Mongolic languages all descend from MM or varieties close- ly related to it (see, e.g., Poppe 1955, Svantesson et al. 2005, Nugteren 2011). They can be divided into 1. Southern Mongolic (SMc) in Amdo; 2. Central Mongolic (CMc) in Greater Mongolia & Kalmykia; 3. Dagur in Manchuria and Tarbagatai and 4. the possibly extinct Moghol in Afghani- stan. 4 SMc consists of Eastern Yellow Uyghur (while Western Yellow Uy- ghur is a Turkic language in close contact to its eastern neighbor) and the Shirongolic languages. According to Nugteren (2011), the latter can be di- vided into a branch consisting of Bonan (=Baoan), Santa (=Dongxiang) and Kangjia and the Monguor branch, which consists of Mangghuer and Mongghul, the latter subdivided into several dialects. As for CMc, Janhunen (2006) suggests it can be divided into the major varieties Khalkha -Chakhar (Central Mongolian), Khorchin -Kharchin (Eastern Mongolian), Oirat (in- cluding Kalmyk) and Buryat, as well as the smaller varieties Ordos and Khamnigan. While Ordos is known for its conservative phonotactics, it has also been classified as part of Khalkha-Chakhar (Luvsanvandan 1959). Ac- cording to my own (limited) familiarity with this dialect, I’d expect it to exhibit the features of the Khalkha-Chakhar group with some influence from Oirat and Western Tümet (Khorchin-Kharchin) but not an aspectual system as distinct as the major varieties. No assessment of the Khamnigan verbal system is possible on the basis of currently available materials (see refer- ences in Janhunen 2005 and Yu 2011). CMc constitutes a dialect continuum. Speakers of the most progressive subvarieties of Khorchin and Khalkha who have not had any previous exposure to the other variety nor to the more simi- lar standard languages should have a fair chance of communicating by and large successfully on issues not related to modern culture or infrastructure when using careful, slow articulation. Speakers of Kalmyk (Oirat) from the westernmost end and Mongoljin (Khorchin-Kharchin) from the very eastern
3 Using a single cover term for what is probably a wide array of varieties might not be appro- priate, but a sustained linguistic attempt to classify the linguistic differences of various Clas- sical Mongolian sources along temporal and spatial lines has never been undertaken. In this paper, this term only works because I refrain from making any taxonomical statement about it. 4 Amdo, Greater Mongolia and Manchuria are all ethno-culturally definable macroareas with- out exact political counterparts. Amdo includes all of Höhnuur (“Qinghai”) and parts of Si- chuan and Gansu, China. Greater Mongolia includes all of the Mongolian state, Inner Mon- golia (China), Buryatia (Russia), substantial parts of Xinjiang (China) and even smaller areas in Höhnuur, the Chinese provinces bordering Inner Mongolia, Tuva (Russia) and Kyrgyzistan. Manchuria consists of the provinces Heilongjiang, Jilin and Liaoning, the subdivisions Hulunbuir, Hinggan, Tongliao, and Chifeng of Inner Mongolia, all in China, and Primorskij Kraj, southern Khabarovskij Kraj, the Jewish Autonomous Oblast’, Amur Oblast’ and Sakha- lin in Russia. Tarbagatai ‘having marmots’ (rendered as WFKpQJ ‘ta -city’ in Mandarin) is an administrative unit within Ili, Xinjiang, China.
3 end of the dialect continuum cannot successfully communicate on all but the simplest issues without some weeks of exposure. The overall differences between these extremes might be comparable to the difference between High German and Danish. 2. Terminology
In this section, I will give short definitions of some of the terms that I use in my thesis. As these definitions are mainly intended to help readers who are not familiar with the study of tense, aspect and evidentiality or with Mongo- lian studies, I will try to provide fairly simple definitions with no claim to technical precision. This section is thus not intended to replace a thorough familiarity with the specialized literature. In the following, Khalkha and Khorchin forms are given in phonemic WUDQVFULSWLRQ IROORZLQJ %D\DQFRܵWX DQG 6YDQWHVVRQ HW DO U e- spectively, with slight modifications). Suffixes are cited in unrounded forms without advanced tongue root. 5 For MM phonemes, a conventional tran- scription that reflects an older stage of reconstruction is used.6 Capital letters indicate vowel- or consonant-harmonic variants.
2.1 Tense
Tense (basically following Klein 1994) can be defined as the grammatical expression of the relation between the time when a particular statement was uttered and the time to which the utterance refers. Utterance time and refer- ence time may precede, follow or partly include each other in different ways, and some of these can be grammaticalized, e.g., as past, present and future tense forms. If the time of encoding and decoding differ, e.g., when reading a text or watching a video, a speaker may have some freedom as to what time of utterance she assumes. Consequently, a concept such as “present” can be defined narrowly or widely. If (cf. Comrie 1985: 38-9) a speaker re- fers to a time before the utterance time, this doesn’t necessarily mean that she claims that the event itself is over. Tense is not limited to locating two points in time relative to each other but can also express remoteness distinctions . Remoteness may either be defined through clear cut-off points such as between ‘earlier today’ vs. ‘be-
5 However, the assumption that Khalkha vowel harmony is velar in nature is quite common in the literature (see Svantesson 2005: 222 for a partial overview), 6 I am reluctant to reconstruct MM vowels. Ko’s (2011) reconstruction might be preferable to the traditional reconstruction followed by Svantesson et al. (2005), and so far nobody seems to have seriously engaged in Ko’s line of argument. For Early MM consonants, I tend to accept the reconstruction of Svantesson et al. (2005) with oppositions of voiceless aspirated vs. voiceless unaspirated and /h/ for some instances of < ܵ>.
4 fore today’ or merely draw a distinction between recent vs. non-recent (Comrie 1985: 87, relying on Dahl 1984). It has been suggested (Botne 2012) that even the distinction between a “currently relevant time unit” (e.g., earlier today, earlier this month, etc.) vs. “adjoining time units” (on an earli- er day, month, etc.) can be primary. While clear cut-off points don’t exist in Khalkha (Song 1997) or CMc, there have been appeals to more vague no- tions of remoteness such as distal vs. proximal (Binnick 2012).
2.2 Aspectuality
Aspectuality is a cover term for notions related to the inner time structure of events such as (i) the inherent time-structural characteristics of lexical units, (ii) time-structural meanings coded or asserted through morphology, (iii) the bounding potential of definiteness and quantification in noun phrases, (iv) the bounding potential of spatial, temporal and other adverbials, (v) the way in which auxiliaries, serial verbs, etc. modify these time structures, (vi) the contribution of voice and, more in general, the expression of causation, and (vii) the way in which clause combinations affect the interpretation of claus- al time structures (adapted with slight modifications from Sasse 2002: 263). In the following, I will assume a bidimensional model of aspectuality and group (i), (iii), (vi), spatial elements of (iv) and non-flexional parts of (ii) and (v) as “aktionsart,” while flexional parts of (ii) and (v) will be character- ized as “aspect,” The idea behind this division is that aktionsart provides an abstract time structure upon which aspect operators operate. Assuming there are a limited number of aktionsart classes, one doesn’t have to be able to explain how their actional structure is composed in order to describe their interaction with aspect, which is of primary interest in this thesis. Aktionsart can be defined as “the phasal time structure of the event” de- noted by a “verb plus its argument frame” (Sasse 2006: 535), i.e., as the bor- ders, phases, and temporal stability of an event. For instance, in Yesterday, we went to the beach , the expression we GO to the beach contains a phase during which the event takes place (which could be asserted by using the aspectual form were V-ing ) as well as a finite boundary (which could be asserted using the aspectual form V-ed ) (Sasse 2006: 535). The function of aspect is thus to assert (“select,” “focus,” etc.) a certain lexically inherent phase of an event (cf. Breu 2005: 51). Through cognitive processes such as UHLQWHUSUHWDWLRQEDVHGRQJRRGFRQWLQXDWLRQ ĺ]RRPLQJRXWVHULDOL]DWLRQ RUVFDODUDGMXVWPHQW ĺ]RRPLQJLQ VHH&URIW -8), actional inter- pretations may change: HIT the dog may be interpreted as a single punctual event, or it may be reinterpreted as a phase of action consisting of a number of no longer relevant subevents. Aktionsart is a property of the actional phrase (the verb stem, auxiliaries, its arguments and syntactically coded
5 spatial limitations, while time adverbials are not assumed to play a role) (cf. Johanson 2000: chapters 5, 6). While a systematic investigation of aktionsart in Mongolian (along the lines of, e.g., Kiryu 1999, Tatevosov 2002, 2008) would doubtlessly yield additional actional classes, a number of actional distinctions play a role in this thesis. The definitions given below are slightly less differentiated than those by Croft (2012: 53-64), though the inceptive and inchoative aktion- sarten (Breu 2005) are added. My terminology differs slightly from Croft’s. The actional classes are illustrated by short constructed examples from Khalkha: States are events that lack perceivable inner dynamics and, with exceptions that don’t matter here, have temporal duration. Depending on whether an endpoint is indicated, we can draw a distinction between tempo- rary (SDMăU -WҦDLSDL -n happiness-COM COP -DIR ‘is happy,’ DPҨWDU -WѻҦL -n live-PROG - DIR ‘is living (somewhere)’) and undelimited (WҦLUPэƾњэѦ[Xƾ DEM .DIST Mon- golian person ‘she is Mongolian’) states. Inceptive states include a lexicalized inception point which can be asserted by a perfective form ( [LZWҦ -WѻL -n ‘is lying,’ [LZWҦ -WѻH lie-2H.PST ‘lay down, lay (then),’ cf. DPҨWDU -Wѻ e ‘lived,’ but not ‘started living (somewhere)’). Activities have duration as well, but they are more dynamic and thus often require a continuous effort to be sustained. They lack lexicalized borders; thus, a perfective form cannot refer to its past inception or endpoint, but only to the event in its entirety ( WҦэњэѦ -WѻH ‘played’). Prototypically, they are undirected (WҦэњэѦ -WѻL -n ‘is playing’), but there are also directed activities with gradual, incremental change into one direction ( [DѦ҂U -WѻҦL -n ‘is getting hot’). Accomplishments are defined as having both duration and a defined endpoint, upon reaching which, they are completed. If the process they are based on is not cumulative, they may be undirected, but directed accomplishments ( [эӃѦ -ig it-WѻL -n food-ACC eat-PROG -DIR ‘is eating the food’; [эӃѦ -ig it-WѻH ‘ate [i.e., completed eating] the food’) are more common. Inchoative states are a blend of an accomplishment and a lexicalized resultant phase. Progressives can refer to both of their lexicalized phases ( Q҂ӃJW -WѻL -n ‘1. is going into hiding 2. is staying hidden,’ Q҂ӃJW -WѻH ‘hid’). Finally, achievements are punctual events without any notable dura- tion. They can be directed in the sense that they culminate in a non- lexicalized resultant state ( un ъի -WѻH ‘fell’), or they can be undirected = cyclic in the sense that they can immediately be repeated ( WVэW -WѻH ‘hit (once / sever- al times).’ As illustrated with progressive and perfective forms in the last paragraph, aspect markers operate upon aktionsart and assert relevant phases. More generally speaking, aspect markers refer to the attainment or non-attainment of the actional borders or the attainment of resultant stative situations. In the latter case, it is useful to assume a reference time / topic time (thus a time phase) (e.g., Klein 1994) or an orientation point (Johanson 2000) that differs from the event time and possibly also from the here and now. The aspect is imperfective if it refers to the internal time structure of single or a row of
6 repeated events without asserting their borders. It is resultative/perfect if it refers to a state subsequent to the event, prospective if a preparatory state that precedes the event is referred to (Klein 1994: 104), and perfective if it “refer[s] to the attainment of the relevant limit” (Johanson 2000: 135) inher- ent in verbal aktionsart. Moreover, the label “perfective” will also be used for forms that stand in paradigmatic opposition to imperfective and resulta- tive/perfect forms and thus express either boundary-oriented or neutral, but not explicitly perfect or imperfective values (i.e., for -POST -INTRA in Jo- hanson’s terminology). One may try to link (at least some of) the different morpheme types within the resultative/perfect and imperfective forms through the notion of focality : “psychological interest may focus more or less on the situation prevailing at O” (Johanson 2000: 38), i.e., at the refer- ence time. For instance, high-focal forms such as resultatives focus strongly on a resultant situation. This focus is reduced with perfects and, as I would put it, is lost in perfectives. Particular aspect types mentioned in this thesis are the following, ordered by domain. The imperfective domain consists of different variants of im- perfective forms: progressive markers refer to ongoing events, and continu- ative markers refer to events in progress that might be perceived as includ- ing an inception point from which an event has continued up to the time of reference. If a marker is habitual , it expresses multi-occasional regular repe- tition, whether this be due to an acquired habit or not. An attitudinal marker refers to an inclination that sometimes materializes, while a potential marker indicates that an event is only realized if the circumstances are met (Bertinet- to & Lenci 2012). Generic markers characterize a class of entities. It has been suggested that habitual up to generic events form a continuous semantic domain, which in Khorchin is indeed expressed by the single morpheme -na . The perfective domain consists of resultative, perfect and perfective mark- ers which constitute a grammaticalization cline (Bybee et al. 1994: 105). Resultative markers include reference to the completion of a causing event but focus on the resulting state (i.e., when combining with time adverbials), while perfect markers express a resulting state but give more prominence to the attainment of the final border of the event (for a much more detailed, though not necessarily generalizable classification, see Breu 1988). With perfective markers, the time of reference coincides with the time of the at- tainment of the relevant actional boundary or, if such a boundary is absent or contextually dispreferred, with the time of the entire event that is pictured within its boundaries. Events are iterative if they repeat (whether regularly or not) (cf. Bertinetto & Lenci 2012: 854-60) and durative if they have some form of temporal duration (cf. Dowty 1979).
7 2.3 Evidentiality
Evidentiality is most commonly understood as the expression of infor- mation source (Aikhenvald 2004), be it obligatory or not, though some (e.g., Michael 2006) rather define it as the mode of access to information. Plungian (2011: 37-8) proposes that there are three main and a number of subordinate categories of evidentiality that can be grammaticalized in human language: Direct personal access can be visual or via other senses , but even the endophoric perspective which relates to internal states of the asser- tor and access to the event from the viewpoint of a participant exist. Indirect personal access is subdivided into inferential access based on observed results and presumptive access based on general plausible reasoning. Com- mon knowledge can be grouped either as a form of direct or indirect personal knowledge. Indirect non-personal evidence is reported evidence that can either pertain to a specific or an unknown information source. Evidentiality distinctions relevant for CMc are the opposition firsthand (visual + other sensory) vs. non-firsthand (inference, possibly presumptive and hearsay) and direct (synchronic visual, other sensory, inference) vs. indirect (presumptive, hearsay, previous perception). 7 Additionally, the time of acquisition (Tournadre 2008) plays an important role. The choice of a particular evidential form from an inventory is a non-trivial process: the speaker must first classify her actual information source according to a grammatically, culturally and subjectively determined ontology. Then, she has to assess her interactional setting, taking into account cultural conven- tions. This then leads her to adopt an epistemological stance , which in turn serves as the basis for her linguistic choices (Mushin 2001, mainly p. 82). Speakers must “take a stand on how they acquired the information.” That is, while there is an inventory of evidential forms, it is still the speaker’s choice how she shapes the information she is forwarding. A speaker might for ex- ample adopt a personal stance to show that she is forwarding her own ver- sion of the events as she perceived them, or she can rather use a factual stance, implying that either “the information is assumed to be known by anyone in the speech community as general cultural knowledge” or “that the source of information is unimportant to the establishment of the validity of the information” (Mushin 2001: 52, 59, 74). In MM and Oirat, this distinc- tion seems to be of special importance because they have a verbal paradigm with two forms that evoke a particular source of information, while a third, highly frequent form is neutral towards evidentiality and time of acquisition.
7 De Haan (2001) defined direct and indirect differently (see the discussion of Khalkha non- past forms), but his discussion served as a starting point for the delimitation suggested here.
8 2.4 Structural grammatical terms
The morphemes mainly used for the expression of tense, aspect and eviden- tiality in CMc are verbal suffixes that can be divided into roughly three cate- gories. Finite verbal suffixes can only be used in finite predicative position, but not attributively or adverbially. They can be subdivided into two subcat- egories, namely, indicative suffixes related to the expression of tense, evi- dentiality and (within paradigms) aspect on which the discussion will focus and mood suffixes with second, third and first person imperative meanings. Participial suffixes are defined by their ability to occur in attributive posi- tion to a noun. The discussion in this thesis is limited to uses in finite posi- tion or within complex finite predicates, together with the copula pai- ‘be (at/with)’ or the inceptive copula SэѦ - ‘become.’ Converbal suffixes are defined by their ability to link clauses. Some express fairly specific mean- ings like ‘as soon as,’ while a few merely seem to add unspecific semantic notions such as a tendency to imply that two events occur in a sequence or with partial temporal overlap. These later converbal suffixes can lose their ability to mark independent clauses and be restricted to adverbial modifica- tion such as Khorchin -Wѻ or even become fossilized in certain contexts such as spoken Khalkha -ƾ. By convention, they are still called converbal suffixes. Linking converbs can be used in complex finite predicates together with the copula or auxiliaries such as üz - ‘see’ (experiential, conative). Mongolic converbs are never used in attributive position, and uses in finite position such as exist for Khalkha/Khorchin -at are exceptional. The class of finitely usable morphemes thus consists of finite verbal suffixes plus those participi- al and converbal suffixes that have a finite usage. 3. Previous research and methodology
While the morphological forms of CMc are almost all identified and roughly categorized, relatively little is known about specific morphosyntax such as analytic TAE constructions or the TAE systems as such. Leaving aside pa- pers on subsystems, integral attempts to describe most or all of the TA(E) system of a CMc variety were undertaken by Bläsing (1984) for Kalmyk Oirat; Byambasan et al. (1987), Svantesson (1991), Kim (1995), Song (1997) and Brosig (2009) for Khalkha, Cinggeltei (1959) and Chuluu (1998) for Inner Mongolian varieties, as well as Matsuoka (2008) for Khorchin and MM. 8 The problems with these studies are manifold, though. Bläsing didn’t include evidentiality in his analytic model. He only considered a subset of
8 Another source that I dare not evaluate here is Dugarova (1991). She seems to focus on aspect in a strict sense but, in contrast to Matsuoka, contrasts simple and complex forms. She doesn’t address the evidentiality complex.
9 complex constructions of aspectuality in a narrow sense (e.g., excluding grammatically marked continuativity) and only worked with literary texts. He thus missed both the evidential system and the complexity of the aspec- tual system of Oirat. Byambasan et al., Kim, Chuluu and Song generally structured their analysis around finitely usable morphemes, thus missing or ignoring their systematic interaction with aspectual constructions. With the exception of Chuluu, they failed to address the position of -Vъƾ as a past tense marker that contrasts with -Ѧъ and -WѻH . Kim made extensive use of inferior evidence such as isolated, partly constructed sentences. Song went a long way with a textbook corpus, constructed sentences, some elicitation and solid general linguistic background but still introduced some artifacts such as constructed conversational sentences with past tense -w into his analysis. Svantesson, by applying Dahl’s (1985) questionnaire, arrived at correct meanings and structures but generally didn’t provide details on semantic nuances and overall structural complexity. The late Cinggeltei’s consistent application of the distributional method as early as 1959 was utterly admira- ble, but his implicit belief that different Mongolian dialects closely resemble each other in grammar combined with the very limited theoretical apparatus available to a linguist in China at that time rendered many of his explana- tions fuzzy. As an educated speaker of Khorchin with an apparently good command of the Inner Mongolian standard (which is mostly based on Chakhar), Chuluu entertained the same mistaken belief as Cinggeltei – in his case, explicitly – and even took his own intuitions as the sole object of the description. Similarly, Matsuoka’s attempt to investigate Khorchin by work- ing with ad-hoc constructed examples and two presumably highly educated Khorchin informants often didn’t yield valid results. Both scholars accepted mixed Khorchin-Chakhar data that surpasses the language used in class by average middle school teachers of Mongolian (i.e., the Chakhar standard) from Khorchin. Matsuoka’s assumption that tense and aspect can be ana- lyzed separately led to an analysis where simple finite “tense” forms are missing from the overall picture. My own MA thesis (Brosig 2009) clarified the existence and approximate extent of differences between different text genres but failed to solve many problems due to insufficient corpus size, time allocatable to elicitation and the sole reliance on Aikhenvald (2004) for the analysis of evidentiality. To avoid previous methodological shortcomings, I took a number of methodological decisions for the analysis of modern Mongolian languages to try to ensure a consistent analysis:
x Spoken conversation is taken as the primary register of analysis. x A corpus of sufficient size is used to obtain natural examples. x Native speaker intuition guides text analysis. x Modified corpus sentences in their original context rather than constructed sen- tences serve as the basis for elicitation.
10 x A sufficient number of informants ensure that the analyst can understand idio- syncratic answers rather than depend on them and avoids turning the biases of individual informants into results of analysis. x The analysis is based on recurrent morpheme combinations instead of single morphemes so as to detect non-compositional meanings.
Another modification was to add a diachronic and areal dimension in order to arrive at an analysis that is diachronically plausible. Khalkha, as the most prestigious current dialect, and Middle Mongol, as its best-researched ances- tor language, were obvious choices, while Khorchin-Kharchin was chosen as the dialect group most distinct from Khalkha within Inner Mongolia. Khorchin was then chosen over Kharchin due to the much better availability of speakers. The only thing known in advance about TAE in Khorchin was Matsuoka’s analysis - which suggested a greater similarity with Khalkha than actually found - and the obvious dialectal difference in the use of -VΩQ and -Wѻѓ between eastern and western regions of Inner Mongolia mentioned already by Cinggeltei. Language contact was not a factor for the selection of Khorchin. Given this scope, I was forced to make certain concessions:
x With time for only three synchronic analyses, I chose to leave aside Oirat, Bur- yat, and historical texts from the late 16 th century to 1900. x Instead of analyzing aspectuality as a whole, I focused on aspect and neglected aktionsart. x I only analyzed finite declarative predications, even if aspect is also expressed in other syntactic positions such as the complex of attribu- tive/nominalized/negated sentences, in adverbial sentences, in combination with postpositions, and reflected in time adverbials. x I had originally intended to account for different written registers of Khalkha (such as scientific literature, different newspaper genres, fictive and non-fictive narratives, and legal texts) but had to content myself with a spoken corpus con- sisting mostly of dialogue. As for MM, I chose to work with only part of the existing corpus.
During elicitation with modified original sentences that only differ by the TAE forms of finite predicates, most feedback of informants was to some degree linguistically relevant, even if the only difference a given informant identified was the degree of politeness. Still, notions such as duration, repeti- tion, degree of remoteness, degree of certainty and access to information were often more central to elicitation, and after an initial evaluation of the different meanings of a minimal pair by the informant, I would ask about those. Another crucial issue was felicity conditions. Informants were ex- pected to indicate if a certain marker would not fit into the context, and an- swers that described or elaborated on settings in which certain forms would make sense were usually considered valuable. I tried to work with inform- ants with good contextualizing skills that could reflect on situations on a variety of different levels, while informants with lesser imaginative abilities
11 or a tendency to reduce the big picture to just a few social elements were less efficient. Overall, the methodology I used enabled me to obtain fairly detailed data on semantic nuances that, when taken together, provide rather reliable in- formation about the actual meaning of forms. Informant intuition was only used for tasks in which it performs well, namely, for decoding the “meaning ... of a single whole expression,” while the task to abstract from this infor- mation was left to the trained linguist. A relatively great diversity of uses was ensured by the corpus. This is superior to any approach that tries to use informant intuition directly on “the meaning of a closed-class morpheme” or “all the senses of a polysemous morpheme.” It is also superior to translation tasks or grammaticality judgments as one might always run into “forms or constructions that strike one as incorrect or ungrammatical and that one is sure one would not use in speaking but that in fact regularly occur in one’s own fluent colloquial speech” (cf. Talmy n.a. for these accessibility notions). This doesn’t mean that the data collection process could not have been im- proved. The small size of the corpus and its limited number of genres meant that some uses were underrepresented or missing, which might have partly been compensated for by other means. For some easily identifiable situa- tions, such as data on non-visual perception, elicitation with non-linguistic stimuli and the creation of realistic situations in which informants would have been expected to use certain forms might have improved results. For forms with insufficiently established usage patterns, one might have at- tempted to get informants to imagine additional contexts in which these forms could have been used, though such evidence would have needed inde- pendent collaborating evidence. A second round of elicitations after the analysis of Khalkha data was completed would have been helpful but proved impractical due to the sheer extent of the Khalkha TAE system, which re- quired me to use all elicitation time for the first round of elicitation, which was later only complemented by small-scale email/messenger-based elicita- tion. The low number of tokens of complex past forms in the Khalkha cor- pus, which ultimately prevented the assessment of this set of forms, could presumably have been overcome only by a substantial larger corpus or by a somewhat larger corpus created such that reference to recently discovered past events as well as narrative monologues would have been highly fre- quent. 4. The case studies of TAE in Mongolian
In this section, I will give an overview of the four papers that make up my thesis. In 4.1, I will summarize my research on aspect and evidentiality in MM. Section 4.2 gives an overview over my research on tense and aspect in Khorchin. Sections 4.3 and 4.4 contain summaries of the papers that deal
12 with evidentiality, aspect and related notions in the present respective past tense system of Khalkha.
4.1 The aspect-evidentiality system of Middle Mongol
The paper on Middle Mongol is based on a corpus consisting of the "Secret History of the Mongols," which details the life of Cinggis Khan, four bio- graphical texts, one philosophical text and occasional reliance on other sources. It divides the MM TAE system into two general domains, the im- perfective domain and the perfective domain. The imperfective domain comprises three tenses:
i. The single suffix with future meaning is -QU .-mU only refers to future events in conditionals or if followed by the particle je . ii. Present tense is expressed by -yU for generic situations, -mU for progressive and habitual situations, and -n bu-yu as a specialized progressive. iii. The past marker that structurally corresponds to the Present Progressive, -n bü- lüge /aju ћX , has a somewhat wider application to both progressive and habitual events. While there are some suggestive examples, past reference cannot be es- tablished as a function of the suffixes -yU and -mU .-QU bülüge /DMXћX express- es a future development at a reference point in the past given conditions that ei- ther obtained (yielding a past habitual reading) or did not (yielding a past coun- terfactual reading).
There are also some miscellaneous markers such as -i, which is only attested with very few verb roots, and -d, which expresses a plural subject, but no clear temporal or aspectual notions. The perfective domain consists of three past perfective and several perfect and/or resultative markers.
x The past perfective markers are -bA, -lUGA , and -jUGU. As was already de- scribed in great detail by John Street (2009), -lUGA functions as a marker of firsthand information, while -jUGU is used in contexts of inference and hear- say. Still, -lUGA and -jUGU seem to be used only in such firsthand and non- firsthand contexts where the source of information makes a difference to the addressee, while the clearly more frequent suffix -bA seems to be used for un- problematic facts. In complex aspectual markers, including past imperfective ones, only the suffixes -lUGA and -jUGU occur, whereas -bA is excluded. x The participial suffixes -GsAn and -GA are rare on their own but seem to ex- press perfect and resultative notions. The resultative converbal suffix -jU com- bined with a copula in -mU expresses resultative or continuative notions de- pending on the aktionsart of the predicate. To express past reference points, both the participles and the converb can combine with past copula forms.
13 Combinations of participles with the verb bol - ‘become’ refer to entering into a state where – depending on the semantics of the participle -- an event either will or has already come about.
4.2 The tense-aspect system of Khorchin Mongolian
Research on Khorchin was based on six hours of transcribed corpus materi- als from the Khorchin Eastern Central banner collected in 2011 and supple- mented with a small number of transcriptions from other areas and a few examples taken from non-transcribed recordings. Relevant examples from the corpus were analyzed on the basis of native speaker evaluations obtained during approximately 80 hours of elicitation in 2012. The most salient division of the Khorchin tense-aspect system runs be- tween the non-past suffix -na and the past tense suffix -Wѻѓ . They have the short variants -n and -Wѻ , which are required with illocutionary particles and questions; thus, the finite vowel that is most often present in absolute finite declarative contexts seems to correlate with assertive force. It is basically -na and -Wѻѓ that can combine with a number of aspectual constructions. Imperfectivity is mainly expressed by three forms:
i. -na is used for generic and present habitual (though zero is used for non- delimited state predications based on nominals). It also covers future and, in ra- re cases, past potential meaning. ii. -Wѻѓ -na is used for present progressive meanings. It is unrestricted by aktionsart and even used for habitual situations with relevant internal structure, in which case it combines with adverbials such as in uder-t=een xvvr tat-tqE-n day- DAT =RPOSS violin draw-PROG -NPST ‘they are playing the violin every day.’ iii. -Wѻѓ -Wѻѓ functions as a past imperfective. In habitual contexts, it can implicate temporal remoteness vis-à-vis -Wѻѓ .
An alternative to the Progressive is the Resultative -DW SѓӃ-, which yields continuative or iterative interpretations when combined with non-telic predi- cates. With telic predicates, it yields resultative states. A contracted non- resultative form -DWѓ -, though commonly mentioned in the literature, was not attested in the material and only recognized by my oldest informant. The converbal suffix -at by itself has a finite, temporally neutral usage. It ex- presses continuative meanings without a clear final boundary with almost all aktionsarten, while resultative meanings seem mostly restricted to a subset of achievements. Both the Progressive and the Resultative can combine with the enclitic = l to yield an iterative or continuative meaning irrespective of aktionsart imply- ing an unpleasantly long duration. Finally, -nWѻѓ - denotes prospective aspect. There are a number of minor forms:
14 x -la ~ -Oѓ , which appears to be partly homophonous with the converb -Oѓ , refers to a proximate past or future. In all confirmed attested cases, some kind of firsthand evidence seems to exist. x -VΩQ seems to be a kind of perfect marker. Informants tend to reject it as be- longing to the standard language, but its relatively high corpus frequency (1/11 as frequent as -Wѻѓ in declaratives) seems to contradict that. At any rate, it is the only option in past tense content questions. x -i is used in second- and third-person present habitual interrogative contexts. Synchronically, it takes the same form as the Voluntative, but diachronically it might be related to -n plus the interrogative clitic = i. x Khorchin - ҦW ѻΩJ - seems to express a dissatisfied stance.
Evidentiality is not expressed by grammatical means. In contexts of hearsay and, possibly, in order to avoid responsibility, the hearsay / quotative / com- plementizer verb gΩ- can be used (though the details are anything but clear), and inferences tend to combine with the illocutionary particle ba , a loan from Mandarin that indicates a guess and asks the addressee for confirma- tion.
4.3 Aspect and epistemic notions in the present tense system of Khalkha Mongolian
Research on Khalkha was based on a corpus of approximately 10 hours and its subsequent semantic evaluation during 285 hours of elicitation (for all tenses combined) conducted mainly during 2013 in Ulaanbaatar. The Khalkha present tense system is fairly complicated. First, there are five suffixes used in the absolute finite position of finite predicates:
i. -Wъ ի J is used in habitual and generic contexts. It also has a specialized, contex- tually conditioned function in which it can refer to one-time events that don’t fit into the normal course of events. ii. -n on its own expresses potential developments irrespective of tense. Its long variant -n=a apparently differs in politeness. iii. Finite -a on its own is very rare. It seems to express resultativity with some modal shades. iv. Finite -x on its own is rare. It is used for enumerating unrelated events without situating them in time. v. Finite -at is not infrequent but hardly ever combines with other aspect markers. It expresses a resulting or pertaining situation requiring somebody to take ac- tion.
Second, there are a number of aspectual constructions. Two of them are based on converbs:
15 x -WѻSDL - is usually contracted to -WѻDL - or -WѻL - and can combine with verbs irre- spective of aktionsart. It expresses progressive meaning. x -ъWSDL - yields resultative meanings with telic but continuative readings with atelic predications.
There are also some constructions based on participles:
x -Wъ ի JSDL - is only used to combine the core semantics of -Wъ ի J with evidential or past tense values. x -Vъ ի ƾSDL - expresses perfect meaning, while -Vъ ի ƾ on its own instead expresses past semantics. x -x pai - expresses irreal meanings in certain modal contexts. x -x gi-WѻSDL -, a combination of -x, the verb gi - ‘say’ and the Progressive -WѻSDL -, expresses prospective meaning.
The present system is made up of the suffixes that can be used in absolute- finite position either on their own or in combination with the aspectual con- structions. On their own, -n and -Wъ ի J cover habitual and other non-actual situations. If situations are habitual, but require an internal aspectual struc- ture, the aspectual constructions listed above can combine with -Wъ ի J. This doesn’t seem to hold for the Prospective where -Wъ ի J is only attested when referring to a non-normal course of events, while it is potentially ambiguous with the other aspectual forms. In combination with -n or -a, the aspectual forms yield various kinds of ongoing situations, lasting results or current relevance. If the speaker has direct evidence for the event or its result, she uses -n, while Indirect -a is used if this is not the case. Thus, a Perfect such as -sъի ƾSDL -n may be used for an event that is inferred from firsthand evidence or to a witnessed event, while a Progressive in -WѻL -n can only mark a witnessed event, contrasting with a synchronically non-witnessed event in -WѻL -[њ]-a. The rare -Wъ ի J SDL - cannot be taken to express an ongoing situation or result as such but appar- ently refers to the form of evidence currently available for a habitual event.
4.4 Factual vs. evidential? - The past tense forms of spoken Khalkha Mongolian
The basic past tense suffixes in spoken Khalkha are -Vъ ի ƾ,-Ѧъ ,-WѻH and the peripheral -w. The basic opposition is between established knowledge (-Vъ ի ƾ) and non-established (mostly new) knowledge, which is then further differen- tiated into firsthand (-Ѧъ ) and non-firsthand sources (-WѻH ). This adds the factor “time of acquisition” to “source of information.” However, vivid rec- ollection and deferred realization allow for using -Ѧъ and -WѻH , respectively. Additionally, -Ѧъ is used to establish a fictive scenario in discourse. In the corpus, past -Vъ ի ƾ is thrice as frequent as past -Ѧъ and -WѻH combined and due to its opposition to the latter seems to acquire a connotation of factual, relia-
16 ble information. In declaratives, -w accounts for just 0.7% of past tense uses. It is used for events that surprised the speaker in the past. In questions, -WѻH is used to ask the hearer to give an answer based on inference. In self-directed discourse, -Ѧъ is used by a speaker who tries to remember something she once knew, irrespective of whether this knowledge was acquired as firsthand knowledge or not. All past markers have future uses. For an event for which the speaker has sensory or internal evidence (including when the speaker refers to her own intentions), -Ѧъ is fairly common. Clues as to whether a future or past inter- pretation hold are mostly syntactical, but stative aktionsart or the presence of the boundary-actualizing marker -WѻҦ - restrict the interpretation to the past. -Ѧъ can be used in questions about the future in which case the speaker seems to motivate her question on the basis of a presumption based on firsthand evidence. The morphological form of -Ѧъ in such contexts is dif- ferent from the form used in past questions. -WѻH can be used when a future event is inferred, and -Vъ ի ƾ marks it as inevitable. Both are exceedingly rare in future contexts, so that they presumably only work in a salient future con- text. Future -w expresses preventive warnings. 5. From Middle Mongol to Central Mongolic
I will now address systematic changes in the inventories and functions of TAE markers between MM and the modern Central Mongolic varieties Khalkha, Khorchin and Oirat. The comparison is structured as follows: In sections 5.1 and 5.2, finitely usable past respective present forms will be compared. In 5.3 and 5.4, the types and complexity of aspectual construc- tions in finite predicates are discussed. Due to the lack of data, the discussion in 5.4 doesn’t include Oirat. In 5.5, final particles related to evidentiality are discussed. In the tables below, a question mark is used if the meaning or paradigmat- ic status of a certain marker are unclear. The S / C columns indicate that a form or one of its meanings is attested in Simple / Complex predicates.
5.1 Finitely usable past forms
Finitely usable past forms in MM (see table 1 below) constituted a system of three suffixes: -bA was evidentially neutral, -lUGA marked visual (and pos- sibly other sensory) evidence, and -jUGU was used for inferred and hearsay information (Street 2009). The evidential markers were probably used when it was necessary to indicate the source of information, while -bA was used in contexts that allowed for a factual stance. The ratio between -bA and the combined token frequency of the evidentially marked suffixes -lUGA and -
17 jUGU in the largest MM source is about 3:1. However, aspectually complex forms (see 6.3) only allowed for either -lUGA or -jUGU . As -jUGU probably goes back to a resultative in *-jU a-QU and as even -lUGA might possibly be analyzed into two elements, *-lU with unclear meaning and *-GA as a form of the MM resultative participle -GA , no evi- dential system can be reconstructed for Pre-PM. However, if
9 Here and in the following, I often generalize from Kalmyk to Oirat in general. Secondary sources on other Oirat varieties such as those spoken in Xinjiang and the contact varieties of CMc in Amdo, Alshaa and western Mongolia are usually not available with the exception of Orulamjab’s (2013) master thesis on Xinjiang Torgut. Given the contact situation of the re- maining varieties, it is not a priori clear whether extrapolations from Kalmyk are feasible. 10 Bläsing (1984: 94-5) misinterprets some future-referring Khalkha examples with -Ѧъ in a non-future Kalmyk sense, though.
18 thus probably not defendable. 11 -w retained the distributional restriction of MM -bA not to occur in complex aspectual constructions based on partici- ples (Bläsing 1984: 28), and it is rejected by informants in combination with imperfective markers such as the Progressive -GҊD - (Goto 2009: 127-8; Bläsing [1984: 27] contends that -GҊD -v is rare in contrast to -GҊD -la ). Evi- dentially, -w is assumed to be firsthand-like or at least expresses speaker’s confidence (Goto 2009: 129, Orulamjab 2013) and can thus be compared with the factual MM -bA . The status of -Vъ ի Q is less clear. It might be a per- fect (cf. Bläsing 1984: 95-7), but it is infrequent at any rate. In contrast to MM, the structure -Vъ ի QEѓӃ -na does exist (esp. Skribnik & Seesing 2012, also Bläsing 1984: 28, Orulamjab 2013). In (spoken) Khalkha and presumably in Khorchin, the previously factual -bA has turned into a marker used in “modal” contexts. In Khalkha, it was replaced by -Vъ ի ƾ (PM *-sAn (Bese 1971) or *-GsAn ), originally apparently a present perfect that might have retained this function in Khorchin. 12 The suffixes -Ѧъ (< -OXћD ) and -WѻH (< -jiai < -MLћDL , the female singular variant of -jUGU ) more or less retained their former evidential values. All Khalkha past forms have acquired a future usage, albeit it is of rather low frequency for -WѻH and -Vъ ի ƾ. Even the frequency of -WѻH in declarative simple and com- plex forms is very low in my corpus, the ratio being 32( -Vъ ի ƾ):9(-Ѧъ ):2(-WѻH ). There are two explanations for this which might complement each other: 1. The corpus was strongly biased towards firsthand reports. 2. The Direct Per- fect has similar inferential uses as -WѻH and might thus have started to partly replace it. Overall, however, a system of approximately the same structure as MM is still being retained. In Khorchin, -bA was replaced by -Wѻѓ (< -jiai ), which lost its evidential meaning altogether, while -Oѓ (< -la < -OXћD ) is only attested with imminent or recent events. Its precise corpus frequency is hard to assess due to the nearly homophonous converb, but the ratio for -lѓ vs. -Wѻѓ lies between 1:20 and 1:50. -sen (< Proto-CMc *-san ) is used in contexts that might suggest a perfect meaning and in content questions. In positive declaratives, the ratio -sen vs. -Wѻѓ is 1:11, but this contrasts with informant perception that it be-
11 I expect that the difference between -la and -w in literary Kalmyk can be interpreted along lines similar to those suggested by Binnick (2012) for the cognate markers in literary Khal- kha. This would require a partial reversal of the analysis suggested by Bläsing (1984: 93-4). It is likely that their main difference is evidential, a category that can undergo adaptions in contexts that are by and large unsuitable for evidentiality such as novels (see also Oswalt 1986: 33-4 on Kashaya where a complex evidential system is reduced to experiential / quota- tive in narratives). 12 Hsiao (2013) claims that it was -GsAn + copula instead of -GsAn alone that developed into modern finite -Vъ ի ƾ. This is not plausible as MM did not even have the form -GsAn bu-yu , while -GsAn bülüge /bile rather referred to a past resultant state and could semantically not have developed directly into a simple past marker. Of course, it IS significant that a source like the 17 th century Erdeni-yin tobciya HAD a form -GsAn bu-yu , but without any statement about the respective frequency of this form vs. -GsAn over time, little of value can be said.
19 longs to the standard language rather than to the regiolect. Its status in the system thus requires further research. If -Oѓ and -sen have indeed ceased to function as regular past forms, the tripartite MM system would have been replaced by a simpler system centered on -Wѻѓ .
Table 1: Past suffixes in finite position MM / Oirat S C Khalkha S C Khorchin S C -bA FACT .PST x -w past surprise x -w -w preventive x ?warning x past question x x ?past Q x -lUGA 1H.PST x x -Ѧъ 1H.PST x x -lѓ 1H.PST REC x -la 1H.FUT x 1H.FUT IMMI x -jUGU 2H.PST x x -tѻe 2H.PST x x -tѻѓ PST x x -dҊ inferential FUT x -GsAn ?PRES .PRF x -sъի ƾ FACT .PST x x -sΩn ?PRF x x -sъի Q FACT .FUT x
5.2 Finitely usable present forms
In contrast to the past suffixes, the life-cycle of present tense suffixes in CMc (see table 2 below) is rather short. Early MM and Late Eastern MM have -yU ,-mU and -i. The distribution of -i is limited to a few verb stems, and with the possible exception of its combination with the copula, its mean- ing is probably beyond reconstruction. On its own, -yU expresses generic meaning, while the younger -mU(i) (from -n bu-i or -n bu -yu )13 has taken over progressive and habitual meanings. 14 In complex aspectual construc- tions, both suffixes simply express present tense meaning. Neither is retained in CMc. Modern present tense suffixes in Khalkha do all go back to identifiable morphemes of MM and Late Western MM. Progressivity is still expressed morphologically, but it requires two suffixes: -WѻL ->њ ]-ъ /-WѻL -n (< -Wѻ SDL -n
13 This idea is not new ( 0LMLGGRUM 7RܵWDPED\DU -85), though I am not sure where it originated. The claim that -mU(i) is somehow a variant of or derived from a rare nominal suffix -m, which is based on evidence from the MM source Hua-yi yi-yu (e.g., Mo- staert 1977: 74, Rybatzki 2003), fails to account for -U. Reconstructing *-m bU-i/-yu might be an option, but an attested combination such as -n bu-yu (Secret history) or -n bu-i (inscription RISULQFH$UXܵOLQH7|P|UWRJRR LVPRUHSODXVLEOH 14 Hsiao (2013: 1091) seems to imply in her table that the overall text frequency of -yU de- clined from 27.2% of all imperfective/non-past markers in MM to 4.7% in 17 th to 19 th century “Late Mongolian.” However, the materials she compares are not equivalent. Her MM corpus contains a very long religious text with lots of eternal truths all requiring a generic marker, but none of the plentifully available religious sources of “Late Mongolian” was included into her corpus. If one compares texts of similar genres such as the MM Secret History of the Mongols and the “Late Mongolian” Erdeni-yin tobciya, Hsiao’s percentages are 8.4% (32/383) vs. 10.5% (40/384); thus, there is no statistically significant difference.
20 -IPFV .CVB COP -RES .PTCP /-POT < Post-MM (*)-ju bayi-ћD /-nam , cf. MM -ju -RES .CVB , 15 bayi- stop+stand-,-GA -RES .PTCP / LWMM -nam -PROG .PRES ). In such complex forms, both -n and -a signal an ongoing situation. Habitual to generic situa- tions are marked by -Wъ ի J, which is cognate with MM -dAg , a very rare suffix with unclear, but possibly habitual, meaning used on verbs in attributive position. It has been identified with Old Turkic -dOg (Shiotani 2005: 123), which didn’t have a (positive) finite use either, but rather, expressed perfec- tive semantics (Erdal 2004: 293-6).16 The present domain is thus split in a way different from MM. It is even more complicated, as simple -n (< Proto- CMc *-na < LWMM -nam) can mark a situation at any time level including the present, if there is a certain potential for it to occur, and the non-future, if it actually has occurred. This meaning, while not properly habitual, often implicates habituality. In addition to its more complicated aspectual system, Khalkha has introduced present tense evidentiality, which involves distinct meanings for finitely usable suffixes in complex aspectual forms. Thus, -n comes to mark current direct perception and -a (from the MM Resultative Participle -GA ) marks information that is not directly perceived at the time
15 Semantically, both event and reference time of Post-MM bayi-nam and Proto-CMc * pai-na ‘is there’ coincide with the time of utterance, which makes direct observation likely. The status of * pai-[њ]-a is much more complicated. The resultative * -a can tentatively be recon- structed for Proto-CMc as it survived in Khalkha, Oirat and Buryat, but it must have been lost in Proto-Khorchin-Kharchin at the very latest in the early 17 th century before the diversifica- tion into subdialects in the Manchu state took place. Conversely, bayi - had lost its meaning ‘stand’ in the 17 th century chronicles “Lu. Altan tobci,” “Erdeni-yin tobci,” “QXUL\DQJԛXL DOWDQWREFL´DQG³$OWDQTDԛDQ -XWXԛXML ,” in which this meaning – in contrast to its uses for ‘exist, be at’ (n=31-34), ‘stop (doing)’ (n=15) and its auxiliary uses (n=131) – might arguably be assumed in only three cases (Secen 1987: 108-18, see Johanson 1999 for ‘stand’ as a source of copulas in Turkic). pai-[њ]-a could thus have only expressed a meaning like ‘is there (as a result of having stopped)’ if it had been conventionalized before the 17 th century (in which case reference and event time wouldn’t coincide). The grammaticalization of pre- sent evidentiality, on the other hand, must have taken place later than the 17 th century if it was contact-induced from Amdo Tibetan (see section 7). If so, it would be more likely that a non- resultative, possibly stative meaning was the input for the grammaticalization process. As it is resultatives (Bybee et al. 1994: 105) or perfects (Tatevosov 2001) that are assumed to serve as input for the grammaticalization of inferential perfectivity, this would seem less likely. It is hoped that a precise study of existing sources could clarify this confusing state of affairs. 16 Shiotani (2005: 124-8) is not justified when he reconstructs PM *-dAg as a participle. 1. MM -dAg is not attested with direct objects. Thus, there is only evidence for a derivational, adjective-forming suffix. The absence of any reflexes of *-dAg in the Moghol branch (Weiers 1972) and the Shirongolic subbranch of SMc as well as the mere existence of adjectives in -GJDӃU (*-dAg plus instrumental case) in Dagur (Shiotani 2005) don’t strengthen his point. A participle -dag only occurs in Eastern Yellow Uyghur and CMc (Shiotani 2005), including Manchurian Khamnigan (Janhunen 2005: 43). However, Eastern Yellow Uyghur differs from the other SMc languages precisely due to its contact with CMc (cf. Nugteren 2003: 265). Therefore, -dag must have developed into a participle either in Proto-CMc or in one of the CMc languages that also form a sprachbund, but no later than 1630 when the Kalmyk Khan- ate was founded.
21 of speech if attached to the CMc copula pai- ‘be.’ These evidential catego- ries CAN be applied to -Wъ ի J, but this is rare, and they cannot be applied to potential -n. The characteristics of Oirat lie in-between MM and Khalkha. The status of -na and -Gъ ի J seems to be by and large identical to Khalkha (cf. Bläsing 1984: 73, 77-80, Ebert 1999: 324-7), although the suffix -x (< -QU ) has re- tained its future uses (cf. Orulamjab 2013, see also Bläsing 1984: 15), so that a detailed differentiation of the usage domains of -na vs. -x is necessary. Even mono-event past uses of -Gъ ի J might be attested (see Bläsing 1984: 69- 70 who is not right in understanding this example as imperfective). Progres- sive notions can be expressed both by the analytic -GҊEѓӃ -Qъ /jov-Qъ(EѓӃ - COP < ‘stand’, jov- ‘go’) and the more frequent synthetic -GҊъ -. The synthetic form seems to allow for wider imperfective uses than the more strictly pro- gressive analytical form (Ebert 1999: 328-31). If true, this would resemble the difference between MM -mU and -n bu-yu . There are two additional forms: the rare -a refers to a concrete event that started before topic time but covers it, expressing a continuative meaning (Bläsing 1984: 76, see also Ebert 1999: 331-2): Hotn-a uls unt-a ‘Die Bevölkerung der Stadt schlief noch’ (Bläsing 1984: 70-1), i.e., ‘The inhabitants of the city were still sleep- ing.’ 17 In contrast, -DG EѓӃ -Qъ is presumed to disambiguate aktionsart (of “initio-transformatives,” i.e., inceptive and inchoative states) and indicate iterativity or extended duration in a pre-aspectual sense (Ebert 1999: 333-5). The most important difference to Khalkha is the lack of a present-tense evi- dentiality distinction, as -a has a different status, and complex predicate forms like *-GҊ EѓӃ -[h]-ъ are absent (Bläsing 1984 for Kalmyk, Orulamjab 2013 for Xinjiang, Yu. Tsendee, p.c., 2014-04-24, for Oirat in western Mon- golia and historical Oirat sources in Clear script). The Khorchin system, in turn, has become simpler. The LWMM Progres- sive -nam has developed into a habitual-generic-future marker -na , while present progressive meanings are expressed by -Wѻѓ -na which is cognate with the Khalkha form. No productive reflexes of -GA exist in Khorchin- .KDUFKLQ VHH%D\DQFRܵWXIRU.KRUFKLQSURSHU6RGXEDܵDWXUIRU Kharchin, Mönggüngerel 1998 for Naiman), thus there is no evidential con- trast. A participle -tΩJ is reported to have existed in 1950s Khorchin (Bayan- FRܵWX -8) for occasional but certain [=regular] events, but it is no longer used. A finite, non-tensed suffix shared by Khalkha and Khorchin is -at . It is re- stricted to simple constructions and thus is not fully integrated into the TAE
17 This sentence illustrates a considerable problem within Bläsing’s (1984) data. Literary narrative texts allow for a very wide imperfective past use of Khalkha -n, and Kalmyk texts allow for -a,-na and -GҊъ -na . However, this tells us nothing about the probably more restricted use of such forms in past imperfective contexts in actual spoken language.
22 system. In CMc, it moreover links clauses and marks the non-final element in complex predicates. The cognate MM -GAd is restricted to clause linking.
Table 2: Present suffixes in finite position MM S C Khalkha S C Khorchin S C -yU GEN x PRES x -mU PROG /HAB x PRES x -i PRES &? x x -nAm PROG .PRES x -n POT x -nъ PRES x x DIR .PRES x FUT x -GA ?RES .PRES x [њ]-ъ RES .PRES x ¬DIR .PRES x -dAg V>ADJ x -tъի J HAB ./ GEN . x x unexpected x x
5.3 Aspectual constructions in finite predicates
All CMc varieties make use of analytical constructions to express aspect. This means that a participle or converb is connected with an auxiliary to yield a mostly aspectual meaning. In MM, such constructions were possible on three levels: converb plus copula, participle plus copula, and participle plus the dynamic copula bol - ‘become.’ As table 3 (Bläsing 1984: 28, Ben- zing 1985: 25 for Oirat) shows, the same number of converbal constructions is present in MM and all three CMc varieties. However, Khorchin only re- tains the older analytic version -Wѻ S ѓӃ - of the CMc suffix -Wѻѓ - if the (for- mer) focus clitic = l is attached to -Wѻ . MM has two participle-based devices to express irreal past possibility, past habituality and past perfect notions, while Khalkha and Oirat express more or less the same notions with three suffixes. Oirat, in addition, has a rare participle-based continuative form. Khorchin, in contrast, has lost this predication type entirely. MM, Khalkha and Oirat can also express the notion of entering into a state where an event would be likely to come about (cf. Song 1997: 299-300) or where the participant already experiences the resultant state and expresses a certain stance towards it (Kullmann & Tserenpil 1996: 204). Some related constructions have been analyzed as expressing evidentiality in Oirat (Skrib- nik & Seesing 2014). Entering into a state in which a certain habit obtains can be expressed in Khalkha (Song 1997: 302-3) and presumably in Oirat. It is not attested among the few surviving instances of MM -QU bol -. Prospectives going back to * ge - ‘say’ preceded by a future suffix have been innovated in all of CMc. In Khalkha, this suffix is -x, while it is -n in Khorchin. In Oirat, both varieties may exist: Orulamjab (2013, section on
23 -Ҟљ Ӄ -) presents an example sentence with a prospective form in -n for Xin- jiang, while Skribnik & Seesing (2014: 151) give -x for prospective meaning and -n (they assume it to be the “modal converb,” but finite -n could look the same) for an event that nearly took place.
Table 3: Functionally similar aspect constructions 18 MM Khalkha Oirat Khorchin -n bu- progressive -tѻi- -dҊa- -tѻѓ - progressive -tѻ pъL - -dҊ bѓӃ/jow- -jU a- ?progressive -ъWSъL - -ad b ѓӃ- -ъWSѓӃ - continuative, resultative resultative -QU potential fut. at -x p ъL - -x b ѓӃ- irreal future at past RT bu- past RT obligation -GsAn tensed perfect -sъի ƾ -sъի QEѓӃ - tensed perfect a- pъL - -tъի JSъL - -dъի JEѓӃ - tensed habitual -a b ѓӃ- tensed continuative -QU inceptive pro- -x p эѦ - -x bol- deferred future bol- spective -GsAn ?inceptive per- -sъի ƾ -sъի QERO - inceptive perfect bol- fect pэѦ - -tъի J -dъի JERO - inceptive habitual pэѦ - -x gi-tѻi- -x/-n gi- -n-tѻѓ - prospective dҊa-
5.4 Possible combinations
The constructions discussed in sections 5.1 to 5.3 are building blocks that can combine with each other with certain restrictions. To avoid extensive detail, I will look at the length of maximal predicates only. Constructions that are based on PTCP SэѦ -, -x gi-WѻL - or express modal meanings (other than evidential modality, at any rate) are excluded from the discussion. The tables below are compiled from corpus data (MM, Khorchin) and Google searches (Khalkha):
18 Of the two copula verbs of MM, I am listing only the more common for each construction, even though the other copula is attested in some cases. For the purpose of this table, Oirat bil ѓ is counted as an instance of bѓӃ-. RT is short for “reference time.”
24 Table 4: MM and Khorchin complex TAE forms stem converb participle finite attested in +copula +copula suffix V- -n/-jU a- -QU bü- -lUGA Secret History, e.g. §§195,216 -n a- -GsAn a- -jUGU Sino-Mongol Inscription 1362 -ъWSѓӃ - -Wѻѓ - (-WѻSѓӃ -) -Wѻѓ Khorchin
Table 5: Examples for Khalkha complex TAE form patterns stem cvb/ptcp ptcp/cvb ptcp finite suffix + pai- + pai- + pai- V- -ъW /-Wѻ -Wъ ի J -Vъ ի ƾ -n/-ъ/-WѻH /-Ѧъ -ъW /-Wѻ /-Wъ ի J -Vъ ի ƾ -Wъ ի J -n/-ъ -ъW -Wѻ -Vъ ի ƾ -n/-ъ -ъW /-Vъ ի ƾ -Wѻ -Wъ ի J -n/-ъ/-WѻH /-Ѧъ /-Vъ ի ƾ
In MM and Khorchin, the number of attested complex aspectual construc- tions is rather small, but so is the corpus. The two MM constructions that contain two copulas consist of an initial converb followed by a participle and a finite verb in the past. The only complex Khorchin form with two copulas attested ends in a past form, too. As participle-based constructions are absent from Khorchin, it contains two converbs. The status of this construction as consisting of three words can be disputed, as the Khorchin Progressive is obligatorily contracted in the absence of the clitic = l. Complex predicates in Khalkha with three copulas are quite rare. Except for combinations with initial -ъW , token frequencies never surpassed n=4. Table 5 contains a few plausible overgeneralizations on the attested data, while a few structurally strange items with n=1 were disregarded. 19 One can point out a number of general principles: converbs in -ъW are the most com- mon initial element, and they never occupy any other position. 20 The con- verbal suffix -Wѻ rarely ever tolerates an element other than -ъWSDL - preceding it, though a very few instances of -Vъ ի ƾSDL -WѻSDL - are attested. The order of the participial suffixes -Wъ ի J and -Vъ ի ƾ isn’t crucial, and both may occur within one predication twice. The most common final elements are the present evi- dential suffixes -n and -ъ, while unclear restrictions obtain for the past suf- fixes -WѻH /-Ѧъ /-Vъ ի ƾ. Thus, the integration of the elements -Wъ ի J with its ability to integrate perfect, continuative and progressive aspect forms into the scope
19 The following data (with a complete, contextualized example each) was retrieved at 2014- 04-21. No examples besides those listed below were found: -aad bai-j bai-san bai-na /-aa /- dag, -aad /-j bai-dag bai-san bai-na /-aa, -dag bai-dag bai-san bai-na lee /-aa, -aad bai-dag bai-san bai-jee, -aad /-j bai-san bai-dag bai-na /-aa, -dag bai-san bai-dag bai-na, -san bai-dag bai-san bai-na, -aad bai-j bai-dag bai-jee /-laa /-san /-aa, -san bai-j bai-dag bai-laa /-aa . 20 -sъի ƾSъLњ -ъW is a fully acceptable non-finite predicate, though, and it is also the only com- plex aspectual form in which -ъW is used finitely.
25 of habituality, its ability to express non-habitual notions in some contexts, and the advent of the evidentiality opposition has greatly increased the max- imal complexity of Khalkha predicates.
5.5 Evidential particles
MM is not known to have had evidential particles, and no particular hearsay uses are reported for the quotative verb ke(m)e - ‘say’ (Street 2013). In Khal- kha, gi - (< *ge- /DWH00>Nހԥޝ -@ (DUO\00>NހԥPԥ -]) is used as a quota- tive device but also features in a large number of constructions, some of which are related to stance. It is uncertain whether a hearsay meaning can be established. The Khorchin cognate JΩ - can be used in hearsay contexts, no- tably in the form JΩQ Ω , but its use is clearly not obligatory. In Oirat, the combination -GҊ gi-Qѓ 2h.pst say-pot is used for hearsay (Skribnik & Seesing 2014: 162). Khalkha has a particle SLѦH (< MM bü-lüge COP (<*‘stop+stand’)- 1H.PST ) to express recollection (Brosig 2012), and written Khalkha ъWѻ H < MM a-MXћX COP (<‘live’)-2H.PST expresses hearsay and possibly inference. The formal V҂Uъ ի J-WъL information/news-COM is used to express hearsay (Mönh- Amgalan 1998). In Oirat, sand Ҋ > GҊ might just mark evidentiality, while ELOѓ retains its status as a tensed copula (see 6.1). In Kalmyk newspaper texts, -VQELOѓ is used for reliable secondhand (not thirdhand) knowledge; for uncertain inference, a participle followed by the noun EѓӃ dl ‘situation’ either in the comitative case -Wѓ 21 or with the existential negation marker uga is used (Skribnik & Seesing 2014: 158, 162). 6. No proof required: Khorchin in Western Manchuria
In this section, I will try to provide a concise areal explanation for the changes that Khorchin underwent in its development from MM. I will start by sketching the migration of the Khorchin population and their current areal and sociolinguistic situation, then briefly address some general changes in Khorchin grammar and finally turn to structural parallels in the TAE systems of Khorchin and Mandarin. The Khorchin tribe, originally the bearers (-cin being a suffix for profes- sions) of a certain kind of quiver called qor , originated at the banks of the Onon river but moved to Hulunbuir in the first half of the 15 th century (Se- FHQEDܵDWXUHWDO 22 In the 1530s, a group of Khorchin moved to
21 Mönh-Amgalan (1998: 155) also lists this form, but final predicates with -PTCP SъLWъի Ѧ-WъL are extremely rare on the internet. SъLWъի ѦWъL is more commonly preceded by adjectives and used attributively. 22 In contrast, Janhunen (1996: 58, 123) claims that the expansion of the Khorchin into Hulunbuir is a secondary phenomenon at the end of the 19th century. Neither source is partic-
26 the lower Nonni basin, into the current Fùyù county just northeast of Qiqi- KDHU 7XOܵDܵXUL DQG 6RGX -DQKXQHQ 5), which brought them into contact with Solon (Tungusic) and Dagur speakers and enabled them to trade with the Han of the Ming state. After becoming Manchu sub- jects in 1635, the Nonni Khorchin were moved southwest to their current ORFDWLRQ 7XOܵDܵXUL DQG Sodu 2008: 22-3) in the Jirim league (since 1999 called Tongliao city) 23 and the Hinggan league of Inner Mongolia. 24 There are Buryat, Ewenki and Dagur in the north, Khalkha and Baarin in the west, Kharchin in the south and Manchu (who might still have transmitted their language until the early 20 th century, cf. Gorelova 2002: 28-9) in the east. In the Manchu Empire, the Khorchin became subject to strict divide and con- quer politics that forbid Mongolians to cross the boundaries of banners (the administrative unit below leagues), leading to separate linguistic develop- PHQWVZLWKLQWKHVHEDQQHUV 7XOܵDܵXULDQG6RGX 7KHRQO\QRWDEOH contact group was Han farmers. While the Manchu court from 1668 onwards wanted to keep Mongolians apart from Han (presumably to retain them as a reserve army) and took various measures to that effect (such as establishing Mongolian as the only regional language of administration, forbidding Mon- golians to learn Mandarin and prohibiting intermarriage), the high popula- tion density in the Han mainland that developed during the stable Manchu rule, intermingled with occasional famines due to natural disaster, gave Han farmers a strong incentive to resettle in the sparsely populated north. The local Mongolian nobility and administration could collect more taxes (and bribes) from Han farmers than from Mongolian herders, so they resented the segregationist politics of the central government. As a result, so much land was rented or sold to Han settlers that Mongolians in some parts didn’t have sufficient pastures left and were forced to become [semi-]agrarian them- selves. After Russia claimed northern Manchuria, the Manchu government tried to prevent its further expansion to the south by unrestrained colonializa- tion of the borderlands from 1878 onwards (Tsai 1983: 13-73). At the eve of the Japanese conquest in 1929, two thirds of the historical Eastern Central banner in Jirim (which then included substantial parts of the modern Eastern %DFNEDQQHU ZHUHIDUPLQJODQG 7XOܵDܵXULDQG6RGX 25 With the ularly clear about the evidence used for these claims. In any case, in order to move to the lower Nonni basin, the Khorchin would have had to move through Hulunbuir. 23 While the Chinese government has reclassified most immediate administrative subdivisions of Inner Mongolia as “cities” to change their legal status, their infrastructure is rural. 24 In detail, Khorchin is spoken in the following areas: Khorchin Eastern Central banner (~ darqan qosi ܵu), Khorchin Eastern Back banner (~ büwe wang qosi ܵu), Khorchin Western Central banner (~ Tüsheed), Khorchin Western Front banner (~ Jasa ܵtu), Jarud banner (origi- nally not Jirim), Jalaid banner, Dörbed Mongol Autonomous County in Heilongjiang, Qian Gorlos Mongol Autonomous County in Jilin (Bayancogtu 2002: 2) and parts of Hulunbuir 6RGX -DODLG'|UEHGDQG*RUORVPLJKWIRUPDVXEEUDQFK -LƗQJ -3). 25 This is not to say that most of the farmers were Han. To take the current Eastern Central banner as an example, there are 6 townships with substantial Han populations and 6 Han
27 surge of Han nationalism during the Cultural Revolution, the replacement of husbandry with agriculture as main form of subsistence was furthered even more (cf. Sneath 2000: 102-125). Ethnically, Mongolians in Jirim and Hinggan in 2000 remained 45% re- spective 41% of the overall population. 26 These figures don’t directly trans- late into speakers of Khorchin, though. Many urban ethnic Mongolians speak only Mandarin 27 and children of mixed parents (or brought up by mixed grandparents) are counted as Mongolians in spite of their invariably weak command of Mongolian. Many Han have changed their ethnicity to “Mongolian” (which they are entitled to do if they have one distant Mongo- lian ancestor) to gain minority privileges such as not being subject to the one-child policy. Conversely, some of the Han who ended up as farmers in Mongolian villages have a firm command of Khorchin. I would estimate that at least 25% of the overall population of the Khorchin area are proficient in Khorchin. In the western part of the Eastern Central Banner (as well as in Hinggan, FI-LƗQJ -6), a second important group of immigrants are Kharchin (Tümet and Mongoljin) speakers, creating isoglosses such as west- HUQXԥޝYVHDVWHUQԥޝLQZRUGVOLNH XΩӃ[ vs. ΩӃ[ µIDW¶ 7XOܵDܵXULDQG6RGX 2008: 104, 114, 156). As the Kharchin were the first Mongolians to come under strong Han influence, they might have indirectly increased the influ- ence of Mandarin on Khorchin. With very few exceptions among the oldest generation, all contemporary Khorchin speakers in the Eastern Central ban- ner appear to have some command of Mandarin, but proficiency varies. Among the informants and recorded speakers from 2012, 17 farmers, ele- mentary school teachers, shopkeepers, etc. reported a poor command of Mandarin, 19 an acceptable command and 4 a good command. Middle school teachers reported at least an acceptable command of Mandarin, and all officials and university graduates reported having a good command. 28 Mandarin contact seems to have influenced discernible subparts of Khorchin, while others seem unaffected. The percentage of Mandarin loans in Khorchin is substantial, but not overwhelming: Sodu (2013: 83), when communities, but these contrast with 23 Mongolian communities and 4 “husbandry areas,” which do contain Han and mixed villages but are dominated by Mongolian villages. From an overall population of 546,560 in 1998, 370,625 were registered as ethnic Mongolians (Tul ܵaܵuri and Sodu 2008: 101-3). 26 Chinese census 2000, as cited on https://de.wikipedia.org/wiki/Tongliao and https://en.wikipedia.org/wiki/Hinggan_League , retrieved 2014-06-09. 27 For instance, To ܵtambayar (2007) reported the results of a questionnaire sent out to 180 ethnic Mongolians in government agencies working in the actual city Tongliao in 2005. Out of 148 respondents, 141 claimed to have an active command of Mongolian; 135 provided information about their children: 78 (54 of whom attended Mongolian schools) were reported to have an active command, while 63 (57 of whom attended Mandarin schools) were reported to have a passive or no command. 28 I obtained brief self-evaluations of their command of Mandarin from about 3/5s of my informants in their own words, which I boldly classified here.
28 HYDOXDWLQJ D ZRUG OLVW RI &DܵDQTDGD -291), found that 13% of approximate 5000 Khorchin words were Mandarin loans. Of one-thousand words taken from five randomly selected passages from my corpus, 9% were Mandarin. Vowel harmony for non-phonemic vowels and palatalized conso- nant phonemes were lost (along the lines suggested by Svantesson et al. IRU %DDULQ FI GDWD LQ %D\DQFRܵWX ). Here, Mandarin contact might, but need not be a factor. A certain influence of Mandarin intonation is claimed (Sodu 2013: 82), whereas retroflexes such as in Kharchin 6RGXEDܵDWXU KDYHQRWEHHQDGRSWHG:LWKXSWRFDVHV G e- pending on the a QDO\VLV WKH FDVH V\VWHP RI .KRUFKLQ %D\DQFRܵWX 147- LV DV ODUJH DV WKRVH RI &KDNKDU 6HFHQEDܵDWXU DQG Khalkha, with 8 cognate forms. Taking a superficial look, most other kinds of nominal morphology seem to be retained as well, though the loss of the focus marking clitic = l in most contexts (cf. Street 1985 for MM ele ) is nota- ble. The most substantial changes occurred in the areas with the most immediate interactive relevance. Mongolian terms of address and interlocu- tionary final particles were replaced by Mandarin loans, and clause-final adjectival CMc constructions for epistemic and deontic modality were re- placed by Mandarin modal adverbs. Conventions of politeness and human interaction changed noticeably, e.g., through the loss of everyday greetings. Counting is mostly done in Mandarin. Overall, the culture seems to be switching from a mixed Mongolian-Han pastoralist-farmer culture (as doc- XPHQWHGLQ.UHOEDܵDWXU 8UDQFLPHJZKLFKLVEDVHGRQILHOGZRUN with very old people in the 1980s, Urancimeg, p.c., 2013-01-08) to a pure farmer culture. In the following, I will draw some comparisons between Khorchin and Mandarin TAE. Mandarin Chinese has a middle-sized aspect system, while tense and evidentiality are not grammaticalized. This contrasts with MM, which, in addition to a middle-sized aspect system, had three clearly distin- guishable tenses and a three-level past evidentiality system. Data relevant to the development of non-past imperfectivity in Khorchin is shown in table 6 (Mandarin data from Xiao and McEnery 2004):
Table 6: Non-past imperfectivity in MM, Mandarin and Khorchin MM Mandarin Khorchin future -QU Ø -na generic -yU Ø -na habitual -mU Ø -na progressive -mU, -n bu-yu zài -tѻѓ -na “progressive-stative” -jU a-mu -zhe -at b ѓӃ-na continuative -jU a-mu -xiàqù (-at b ѓӃ-na ) inceptive - -TOiL -
29 While MM carefully differentiates between different low-focal notions such as future, generic and habitual notions, and even Manchu (Gorelova 2002: 293) at least commonly uses a particular future form -ra , Mandarin and Khorchin code these notions in the same way. Notably, the loss of the dedi- cated CMc habitual suffix -WΩJ , to whatever extent it might have been used in earlier stages of Khorchin, is recent. Both Khorchin and Mandarin have pro- gressives that can be compared, although Mandarin has actional restrictions absent in the presumably older Mongolian form. What is termed “progres- sive-stative” here exhibits only limited similarities. Mandarin -zhe is as compatible with states as with activities (Xiao and McEnery 2004, in the following abbreviated as X&M, 189, 208-9) and is overall less suitable to express the situational change than zài (cf. X&M 215). With inceptive and inchoative aktionsart, it is more prone to refer to the phase after the initial punctual change, while zài must refer to the initial phase (X&M 191). On the other hand, the “progressive-stative” MM and Khorchin forms yield a con- tinuative action with activities and a resultant state with accomplishments and achievements. Mandarin -xiàqù , then, appears to be a dedicated means to express resumption. As Mandarin uses two highly frequent and two less frequent forms to express high-focal notions, it is not surprising that Khorchin could retain two such forms. Structurally, MM and Mandarin perfective forms are very different. X&M state that the most common way to express perfectivity is via directional, completive and result-state auxiliary verbs (160-1), 29 but Khorchin hasn’t developed auxiliaries corresponding in usage or frequency to Mandarin wán ‘finish,’ dào ‘manage’ or NƗL ‘open.’ Another very frequent means for refer- ring to short non-durative events in Mandarin is verb reduplication (X&M 150-9). Grammars written in Inner Mongolia generally record a number of aspect-related means for verb reduplication (see, e.g., Secenb DܵDWXU 149-50), but, depending on the converbal suffixes used, they instead express notions such as iterativity, distributivity or temporal relationships between clauses. As indicated by this last point, verb reduplication is usually reported for non-finite predicates. Even entirely unmarked predicates are possible in some contexts (X&M 239). Correspondences to the three MM pasts and one perfect form are given in table 7:
Table 7: Perfectivity in MM, Mandarin, Khorchin MM Mandarin Khorchin simple past -bA Ø -Wѻѓ firsthand past -lUGA complete: le immediate: -lѓ secondhand past -jUGU - - perfect -GsAn - -sΩn
29 In their corpus (X&M 159), such constructions had a token frequency of 1741, contrasting with 1165 for -le .
30 In contrast to Mandarin, Khorchin retains a “neutral” past into which it has generalized the MM secondhand past. For comparison with the MM / Khorchin Perfects, the best candidate might be the Mandarin Experiential, but none of the few attested instances of MM -GsAn and Khorchin -VΩQ look like translation equivalents of guò.30 One form that is as semantically dissimilar as guò and -VΩQ , but might possibly help to account for the semantic change of MM -lUGA into Khorchin -Oѓ , is Mandarin (-)le . None of the past suffixes of MM had a fu- ture use, but Khorchin -Oѓ is known to have developed one. Even in Khalkha, the future use of -ѦD is much more common than the future uses of the other past suffixes. Mandarin (-)le , as such, is of course not a marker of tense but of aspect, and it is even two homophonic morphemes, one used post-verbally and one in sentence-final position. Post-verbal -le expresses “complete” situ- ations without introducing a final (spatial or temporal) endpoint and may, if no other delimitation is present in the overall context, be obligatory, e.g., X sì hou ‘after X died,’ but * X sì . Sentence-final le refers to a currently rele- vant state that has newly come into being at a time of reference (X&M 105- 7; 132-8 synthesizing Li et al. 1982: 22 and Henne et al. 1977: 112):
(1) WƗ FKnj -le fàguó (X&M 134 citing Li 1999) 31 he go-ACTL France “He went to France” (2) \Ư -gè ORWjLSy FKnj -lái, jiàn shì M -gè one-CLF old.woman come-out, see are a.few-CLF MQJFKi dùnshí KXƗQJ-le shén policeman, immediately panic-ACTL expression “An old woman came out. She was scared out of her wits when she found that the visitors were some policemen” (X&M 108)
30 Manchu doesn’t seem to have played any role. It has a simple past -ha that is commonly translated as -bA and -lUGA and a perfect -habi, which is commonly used as translation equivalent for -jUGU, e.g., in the Iledkil šastir , an official history of the relations of the Man- chu with minority peoples of their empire. Mijiddorj (1976: 117-9) probably errs, though, when he suggests that both -habi and -jUGU are secondhand past forms. The meaning of - habi is not entirely clear, being described as “past indefinite” by Gorelova (2002: 290-1, cf. Möllendorf 1892: 9) and as postterminal non-past (i.e., present perfect) by Rentzsch (2012: 852). In Manju-yin ünen magad qauli , first volume (unpublished facsimile, library of the Abteilung für Mongolistik und Tibetstudien, Bonn), which is a relatively freely translated parallel text presenting a mystical origin of the Manchu rulers, translations such as -GsAn bu- yu suggest present perfect meaning, and even alternative translations such as wa-habi (line 59) ‘he has killed [XY, now we must take revenge]’ as ala-OXћD (line 58) ‘he killed [XZ {I have good [here certainly not firsthand] evidence}, so let’s take revenge]’ or te-hebi ‘has settled’ vs. VDћX -mu ‘is living’ (both line 22) point towards a perfect. In contrast to Khorchin, Manchu even had a participle-based pluperfect (Möllendorf 1892: 9) translated into Mongol by the evidentially unmarked form -GsAn bülüge . 31 The only particular glossings are ACTL ‘actual aspect marker’ and COS ‘change-of-state.’
31 (3) WƗ FKnj fàguó le (X&M 134 citing Li 1999) he go France COS “He has gone to France” (4) [Lj\· le (X&M 132 citing Zhào 1968) rain COS “It’s raining now”
Perfective uses similar to (1) and (2) and resultant states from completed events such as (3) seem to refer to the past, and in this sense they can also be expressed (or implicated) by Mongolian -Oѓ , be it for recent or (at least in the case of Khalkha) recalled distant events. There are uses such as (4) which are non-past and don’t refer to resultant states. There is still a certain similar- ity to an imminent future as denoted by the Mongolian equivalent, as it doesn’t matter so much whether or not it is already raining or is only about to if you have to choose your clothing. Overall, several uses of (-)le are rem- iniscent of a past perfective marker. In addition to the past, both forms of Mandarin (-)le can even refer to the future. For -le, this use is pretty restricted as it requires a reference time even after the event in question as in (5) (X&M 116, 121-3), while sentence-final le has fairly broad future applications, either neutral such as (6) or linked to volition such as (7), and at least the latter is easily translated by Khalkha -ѦD and possibly its Khorchin equivalent. Even the explicit use of future adver- bials exemplified by these two sentences is typical for future uses of -ѦD .
(5) PtQJWLƗQEƗ -GLQ sì-míng JƗQMQJ NČQGuQJ tomorrow 8-o’clock, four-CLF policeman surely \MƯQJ OtNƗL -le shàoyáng already leave-ACTL Shaoyang “The four policemen will definitely have left Shaoyang by 8 o’clock tomorrow” (X&M 123) (6) liàn-le ]KČQJ]KČQJ \Ư -nián yáng EQJPtQJ practise-ACTL whole one-year, Yang Bingming kuài FKnjVKƯ le soon finish-apprenticeship COS “Having been trained for a solid year, Yang would soon complete his apprenticeship” (X&M 133) (7) zhè yào GXǀVKR qián \Ư -lì Z΅ quán PL -xià le this pill how-much money one-tablet, I all buy-up COS “How much does each of these pills cost? I’ll buy them up” (X&M 133)
Overall, it is conceivable that the past uses of -le and the present uses of le were perceived as analogous to Mongolian -Oѓ , thus providing an incentive to create future uses of -lѓ analogous to future le .
32 Epistemic auxiliaries in Mandarin might help to complete the picture of epistemological stance in Khorchin. For instance, Mandarin modal auxilia- ries like NƟQGuQJ ‘confirm’ > ‘surely’ and NƟQpQJ ‘may’ > ‘maybe’ are both attested in my Khorchin corpus in an adverbial use, and the Mandarin illocu- tionary / modal particle ba ‘probably’ is commonly used in inferential con- texts. The extent to which such devices might express or implicate reference to evidential notions is not clear. The auxiliary ge - ‘say’ basically seems to be restricted to quotative uses in Khalkha (not treated in my thesis) but ap- pears to include hearsay in Khorchin. In Mandarin, devices such as WƯQJVKXǀ ‘I heard it said’ (with VKXǀ ‘say’ used as a complementizer of WƯQJ ‘hear’) or VKXǀ ‘it is said’ are commonly used sentence-initially to express hearsay, often with an overtone of skepticism, as in (8) (Wáng et al. 2003: 464-72, Yang & Yap 2012):
(8) SK-II, VKXǀ PČLWLƗQ ]K shuì \Ư -gè [LRVKt Q [LƗQJ[uQ ma? SK-II say everyday only sleep one-CLF hour you believe Q [A: ‘Recently, I have often been burning the midnight oil. Acne has appeared on my face. What should I do?’ B: ‘SK-II!’ A:] ‘SK-II. It is said that you only need to sleep for one hour with SK-II; do you believe it?’ (Wáng et al. 2003: 470)
Standard Mandarin doesn’t have a hearsay particle, 32 and if VKXǀ is used sentence-finally in Taiwanese Mandarin, it expresses notions such as coun- terexpectation or emphasis of one’s own opinion (cf. Wáng et al. 2003: 272- 7), possibly going back to postponed Z΅ VKXǀ µ,VD\¶ 6nj 2004: 28-9). The restricted but conventionalized extent to which sentence-initial VKXǀ is used for hearsay marking approximately resembles what is known about its func- tion in Khorchin. For both languages, more extensive data would be wel- come, though. Extensions of Manchu se - ‘say’ don’t seem to have been the subject of any research so far. 7. Mixed evidentiality: Khalkha between Central Asia and Amdo
In its current location, Khalkha Mongolian is only surrounded by other CMc and Turkic varieties (Buryat in the north and northeast, Khorchin in the northeast, east and southeast, Shilingol and Chakhar in the south, Oirat in the southwest and west, and Tuvan in the northwest; tiny Turkic varieties such as Dukhan can be found in the north as well). Remnants of Ewenki speakers are still found in the autonomous Ewenki and Oroqen banners in Hulunbuir
32 Taiwanese Mandarin has de VKXǀ (Yap Foongha, p.c., 2014-04-08), and more southern varieties such as Cantonese have, with wá, a full-fledged hearsay particle (Yap & Ahn).
33 in northeastern Inner Mongolia, as are Khamnigans, a group bilingual in Mongolian and Ewenki. Bilingualism was lost among the Khamnigans of Khentii province, Mongolia. The Khalkha don’t seem to have had any direct contact with the Manchu speakers of central Manchuria and Tungusic speak- ers of eastern Manchuria. Contact with languages such as Kazakh in western Mongolia and Uyghur and Kyrgyz in Xinjiang is mediated by Oirat. Chakhar, which belongs to the same subgroup of CMc as Khalkha, is adja- cent to the Amdo area in the south where local Tibetan, Mandarin (e.g., Wutun), Southern Mongolic and some Turkic varieties form a sprachbund (Slater 2003: 6-9, Janhunen et al. 2008: 21-2). The geographically closest Southern Mongolic variety is Eastern Yellow Uyghur, the only variety of Southern Mongolic that is similar enough to CMc to allow a limited degree of intercommunication. Khalkha and Amdo Tibetan 33 were even in direct contact, as the latter is the language of Mongolian Buddhism. This is signifi- cant as monks are reported to have constituted about one-third of the male Khalkha population at the eve of communist modernization 34 and one-fourth around 1925 (Barkmann 1999: 253). I am not aware of studies that explicitly try to measure the precise influ- ence that Turkic languages exerted on CMc after MM times. Earlier contacts led to strong lexical and structural resemblances that were numerous enough to lead to the Altaic hypothesis (see, e.g., Schönig 2003). As far as I know, non-lexical influence of Tibetan on CMc has only been studied for transla- tions (Higuchi, e.g., 1994). I am not aware of any study that has tried to de- termine the extent of Tibetan loans in CMc, but it would be much easier to conduct, as Tibetan loans only started to enter Mongolian during the Late MM period and had distinctly different sound correspondences then than they had at later times (Atwood 2013). Among personal names, the most common foreign names are from Tibetan, and compound names of Khalkha and Tibetan freely occur (cf. Serjee 2010). Unfortunately, I don’t have any data on bilingualism among 19 th century monks nor on the switch from Mongolian to Tibetan as the language of religious services. The tripartite past tense system of Khalkha (as well as Oirat and presum- ably Buryat) was directly inherited from MM and thus isn’t in need of im- mediate explanation. The extended future uses are, but as I lack both evi- dence for similar areal developments (excluding Mandarin) and cannot ad- dress grammaticalization paths due to the exclusion of Classical Mongolian materials from this study, I won’t be able to say anything more on them here.
33 The language used in contemporary monasteries in Mongolia and Inner Mongolia is Amdo Tibetan (Gow-a, a doctoral student in musical sciences from the Central University of Nation- alities who works on recitation in Mongolian monasteries, p.c., summer 2013), as is the lan- guage from which Tibetan loan words were taken into Mongolian (Atwood 2013). 34 Federal research division of the Library of Congress (1986-1998): Mongolia: Buddhism. http://countrystudies.us/mongolia/47.htm , retrieved 2014-06-11
34 The tripartite past system, however, fits neatly into its areal context. Periph- eral Turkic languages such as Yakut and Turkish retain a bipartite system based on a firsthand/confirmative/neutral past -DI and a secondhand / non- confirmative past in -0,ù that structurally resembles Old Turkic. But Central Asiatic Turkic languages such as Kazakh, Kyrgyz, Uyghur and Uzbek have developed a tripartite system based on -DI (as in Old Turkic), -GAN (from a perfect) and -IP (from a perfective converb) (Johanson 2003, Straughn 2011: 164-9). Like in Khalkha, these systems don’t seem to be associated with three different values for source of information. Straughn (2011: 61-82) pro- poses to analyze the Kazakh forms in terms of confirmativity and definite- ness: -DI is used for events that the speaker can vouch for, often but not nec- essarily on the basis of firsthand knowledge, and which took place at a defi- nite (identifiable and clear-cut) time in the past. Confirmativity outranks definiteness:
(9) Nazarbayev 4DUD÷DQGwPHWDOOXUJL\D]DZwW -ï-nda žumïs iste-di. NAME PLACE metallurgy factory-3-LOC work work-PST .3 SG ‘Nazarbayev worked in a Karagandy metallurgy factory.’ (Straughn 2011: 64) (10) Azi Nomhon dalai-n orn-ii ediin_zasg-iin hamt-iin Asia Pacific ocean-GEN country-GEN economy-GEN together-GEN ajillagaan-ii baiguullaga buyuu APEC-iin deed hemjeen-ii work-GEN organization or APEC-GEN upper measure-GEN uulzalt öchigdör Pyerug-iin niislel Lima-d öndörlö-löö . meeting yesterday Peru-GEN capital Lima-DAT end-1H.PST ‘A summit of the Asia-Pacific Economic Cooperation or APEC ended yesterday in the Peruvian capital Lima.’ (Brosig 2009: 133-4) (11) Keše VD÷DW 3-’te düken-ge bar-dï-m. yesterday hour 3-LOC store-DAT go-PST -1SG ‘Yesterday at 3:00 I went to the store.’ (Straughn 2011: 69) (12) Älginde ÷DQD bul kitap-tï oqï-p qal-dï-m. just now this book-ACC read-CVB PFV -PAST-1SG ‘I’ve just now finished reading this book.’ (Straughn 2011: 69)
The writer doesn’t have firsthand knowledge of the event in (9), which in- volves Kazakhstan’s ruler Nazarbayev, but she is very sure of it. While no such uses of -ѦD were attested in my spoken corpus, the language of Khalkha newspapers is perceived by informants in a similar confirmative way dis- played in examples like (10) in which the probability that a Mongolian re- porter actually attended the event personally is low. (11) illustrates the use of -DI with definite, very specific temporal reference. This is very similar to Binnick’s (2012: 102-7, 162-4) tempting analysis of Khalkha -ѦD as a deictic past, even though he thinks that deictic past forms are necessary to anchor an event in time in the absence of specific time adverbials whose presence would allow for an anaphoric (~indefinite) past tense. In (12), reference is
35 made “to the definite time denoted by the very recent past that requires the use of the simple past” (Straughn 2011: 69). I analyzed this use of Khalkha -ѦD in such a context as recent firsthand perception. If some of the uses in Kazakh are register-specific, the values at play here might be rather similar to Khalkha. The meaning of -IP “is characterized by strong non-confirmative meaning ... manifested as doubt, hearsay, inference, or admirativity” (Straughn 2011: 78) depending on context. It is usable in common inferential contexts, e.g., to the illness of an unhealthy-looking friend on the basis of her outer appear- ance, with quotative and hearsay information, unobservable first person events such as falling asleep, forgetting, even visual information from dreams or if one wants to avoid taking responsibility. Given particular social conventions, the core meaning of the form itself might be non-firsthand. -GAN is described as neutral as far as the commitment of the speaker is concerned. It is also aspectually different in being temporally indefinite, a usage, e.g., present in experiential uses. Without any statistical data or more concrete information on how speakers choose a marker given a certain epis- temological situation, the status of Kazakh -GAN vs. -DI is hard to assess. It is reminiscent of Khalkha -Vъƾ , but maybe more of Oirat -w and its MM cognate. The overall similarity of the Khalkha and Kazakh past tense system points to a contact-induced phenomenon, even if the particular rationaliza- tion of certain uses by new generations of learners might have drifted apart a little during more recent times of presumably less intensive contact. Incidentally, even Amdo Tibetan has a tripartite division of forms that are perfective [at least when used on their own, and thus also non-present]. It distinguishes (i) = QΩ /Ø for volitional participation of the speaker (or, in questions, the addressee); (ii) = WҦ for firsthand evidence of events which the speaker didn’t purposefully perform herself (including uncontrolled events such as loving, smelling, non-deliberate sneezing or actions in dreams); and (iii) = ]ΩJ for inference or hearsay regarding events that the speaker didn’t witness at all (Sun 1993: 956-73 for the Mdzo-dge dialect, see Haller 2004: 136-49 for a similar division in the Themchen dialect). This system doesn’t resemble Khalkha. Some SMc languages exhibit a bipartite division even more dissimilar to Khalkha. They roughly draw a line between events into which the assertor (i.e., the speaker in declaratives and the addressee in ques- tions) was involved herself (cf. Creissels 2008: 12 for this wording) and everything else (Slater 2003: 194-220, Faehndrich 2007, Åkerman 2012, on varieties of Monguor; Fried 2010: 186-93 on Bonan, cf. the distinction be- tween ȕH /ȕD vs. ȕDL in Eastern Yellow Uyghur, Chuluu 1994). The fact that Khalkha has innovated non-past evidentiality is somewhat surprising. After all, in its current location it is only surrounded by Turkic varieties that don’t have it grammaticalized and other CMc varieties from which this change can hardly have originated. Taking a look at CMc first, it
36 can be seen that Khorchin and Oirat don’t systematically mark present evi- dentiality, while it seems to be present in Chakhar and Buryat. Next to -ža bajg-aa as in (13), even forms like -dag bajg-aa (which presumably fit better into encyclopedic style than most other registers) can be found on Buryat Wikipedia (as a corpus), and corresponding forms in -na exist as well.
(13) hüülš-yn žel-üüd-te müren-ej uhan aža[_]üjledberilel-ej last-GEN year-PL -DAT river-GEN water production-GEN boxirdol, xüdöö aža_axuj-n xajagdal gexe_mete xün-ej pollution rural economy-GEN waste etc. human-GEN üjle_ažalalgaan-haa bol-ood jexe boxirdo-žo bajg-aa .35 activity-ABL become-CVB very become.polluted-CVB COP -¬DIR ‘During the last years, its water became much polluted due to indus- trial pollution, agricultural waste and similar human activities.’
Much further towards the northeast, in the Khabarovsk region, an opposition between finite and participial verb forms can be found in Nanai and Ulcha (Malchukov 2000: 449-52) that could be interpreted as expressing evidential distinctions in the present. In nearby Udihe, an evidential contrast arises between Progressives such as (14), which must be firsthand, and the neutral present tense, as in (15), which has neutral imperfective and evidential se- mantics (cf. Nikolaeva and Tolskaya 2001: 251, 262). 36
(14) bi xunazi: tege-we wo:-mi bi:-ni me sister.1SG gown-ACC make-INF be-3SG ‘My sister is making a gown.’ {in a situation in which the speaker has returned home and witnessed the situation he describes} (262) (15) bi xunazi: wo:-ni me sister.1SG make-3SG ‘My sister is sewing.’ (constructed, Irina Nikolaeva, p.c., 2014-03- 10)
Thus, a marker similar to the Khalkha Direct Evidential exists, whereas the Udihe equivalent of the Khalkha Indirect is both evidentially and aspectually unspecific. A firsthand vs. secondhand contrast might be most salient for ongoing one-time situations, while Khalkha also can include habitual but not
35 https://bxr.wikipedia.org/wiki/ ɏԛɯɷ_ɦԛɪɷɧ , retrieved 2014-04-25. I’m using the same abbreviations here as in my papers on Khalkha in this volume. 36 There is also a Habitual (Nikolaeva and Tolskaya 2001: 258-62), which can certainly be used in secondhand contexts, but seems to be acceptable in firsthand contexts as well. While a “normally” (260) reduplicated use of the Habitual Participle can refer to an intensive action in progress with atelic verbs, it is not so clear to me whether the nonreduplicated use in bi xuna- zi: wo:-i bi:-ni me sister.1SG make-PRES .PTCP be-3SG ‘My sister is sewing’ {said when outside his home} (262) is accurately translated as a progressive. If not, this example would just illustrate non-progressive, but not non-firsthand meaning.
37 potential situations. Contemporary speakers of these Tungusic languages are approximately in the same location as in the 15 th century (Janhunen 1996: 30, 92), far beyond the grazing grounds of the easternmost Mongols after the collapse of the Yuan Khanate in 1388. As -a appears to be absent in all of Khorchin-Kharchin, the development of present evidentiality in Khalkha cannot have predated the time when the Khorchin left the Onon River in the 15 th century. 37 Therefore, any direct contact between Mongols and Tunguses from the far eastern regions can be ruled out. There were contacts between the Tungusic Ewenki that formed a new bilingual ethnicity, the Khamnigans, in eastern Mongolia and western Manchuria, but Ewenki is not reported to have pre- sent evidentiality. In the Turkic languages, with the possible exception of Western Yellow Uyghur (mentioned en passant by Johanson 2003: 285) and Salar, present evidentiality was never grammaticalized but is confined to non-obligatory tense-neutral evidential particles:
(16) Keše tüngi smena-da iste-p-ti de, qazir demal-ïp yesterday night shift-LOC work-CPST-3 and now rest-CVB žat-ïr eken. PROG -PRES EVID ‘He worked the night shift yesterday; he must be resting now.’ 38 (17) Yag odoog-oor Terelji-d amar-ch baig-aa aj .39 exact now-INS PLACE -DAT rest-CVB COP -¬DIR IP .HEARSAY ‘[The actress Zhao Wei (...) has come to Mongolia. She (...) says that she wants to travel to historical sites of our country.] Right now she is resting at [the tourist resort] Terelj.’ (18) Qazaq eken-der Kazakh EVID -PL ‘They are (apparently) Kazakhs.’ (Straughn 2011: 99)
In (16) from Kazakh, the particle eken (< är-GAN COP -PST ) marks inference or hearsay but doesn’t contribute to time reference. This is reminiscent of the literary hearsay particle aj (< a-MXћX ), as exemplified in (17). However, the dividing lines within the evidentiality system of Kazakh and spoken Khalkha are different: an example that signals inference on the basis of synchronic firsthand evidence such as (18) (and similar sentences in Yakut such as giv- en in Buder 1989: 92) would be translated into Khalkha with the direct evi- dence form of the copula pai-n rather than pai->њ@ -a.
37 Incidentally, the innovation of present tense evidentiality is more likely to have taken place after the westward migration of the Oirat in the early 17 th century, as this event discontinued the areal continuity of speakers of different CMc dialects. 38 Straughn 2011: 114. CPST ‘converbal past’ (i.e., PST < CVB ), EVID ‘evidential.’ 39 D. Dolgor (news.mn): Jüjigchin Zhao Wei (Huan Ju Geg) mongold irjee, http://www.itismovie.co/2011/07/zhao-wei.html , retrieved 2014-04-28
38 Among the present evidentiality systems that I had a look at, the one that comes closest to Khalkha is found in Amdo Tibetan (all information taken from Sun 1993). In the Progressive ҦNΩMRGa ҦNRG , Amdo Tibetan differ- entiates between (i) evidence about events in which the assertor was in- volved marked with zero, (ii) “perceptible evidence directly present in the immediate speech-act situation” (Sun 1993: 977) marked with ҦNΩ , and (iii) previous direct evidence for synchronically still ongoing events marked with the Perfective Firsthand WҦ in the same manner as used for past progres- sive events:
(19) tæ ƾΩ tюæ QWҦRƾ ҦNΩ jod now I( ERG ) tea drink= PROG AUX ‘I am drinking tea now.’ (Sun 1993: 975) (20) W NҦXWю QWK|ƾ ҦNΩ MRG ҦNΩ ‘He is drinking tea now (I just found out).’ (Sun 1993: 976) (21) tѺaюҦi=k Ω tæ mumu OL ҦNRG WҦ name= ERG now momo make= PROX .AUX =DIR {The speaker, who is not in the kitchen now, saw Bkras-shis making Tibetan momos (...) in the kitchen a short while ago and is reporting what he assumes Bkra-shis must still be doing now} ‘Bkra-shis is now making momos (I saw him do it just now).’ (Sun 1993: 978)
As Khalkha doesn’t differentiate between volitional acts of the assertor and events that the assertor merely observed firsthand, both (19) and (20) can be translated into Khalkha with the Direct Progressive -WѻъL -n. In contrast, WҦ in (21) corresponds to the Indirect form -WѻъL>њ@ -ъ, which is also used with hearsay information for which Amdo Tibetan would use the defective quota- tive verb se together with the indication of the source of information origi- nally used by the cited speaker. As the following examples suggest, an evidentiality distinction similar to the Amdo Tibetan Progressive might also be possible with its firsthand per- fect forms in MRG WҦ :
(22) NҦX mumu zΩg li MRG WҦ he( ERG ) momo some make AUX =DIR .EV ‘He has made some momos (I saw it).’ (Sun 1993: 981) (23) ҦO¡E]Dƾ tюRƾ GX mэ gæ MRG WҦ NAME small=when her( DAT ) love AUX -DIR .EV ‘Blo-bzang loved her when he was small.’ (Sun 1993: 969)
For (22), Sun (1993: 969) suggests that the speaker witnessed the event in its entirety (though I am wondering whether it might even be possible that the speaker saw only the results). In (23), the perfect firsthand form is required to talk about emotions of another person, i.e., to refer to some kind of signs of these inner states that can be sensorily perceived, while the speaker could
39 use simple WҦ if talking about her own emotions. Khalkha Mongolian is less restrictive with inner states; they can receive firsthand marking if the link between the outer appearance, etc. of the referent and her inner state is considered to be straightforward by the speaker, while secondhand marking would be used if the visible signs are more ambivalent. But like in Amdo Tibetan, the Direct Perfect -VъƾSъL -n can both be used for events perceived in their entirety and events deduced from their results. The most obvious difference between the Khalkha dichotomy -n/-a and the Amdo Tibetan system is the greater complexity of the latter and the dif- ferent categorial borders reflected in the apparent lack of schematic transla- tion equivalents. The interesting similarity, on the other hand, lies in the ability of both Amdo Tibetan and Khalkha to express current direct percep- tion and contrast it to other notions. Amdo Tibetan is thus more similar to Khalkha than are the SMc languages, most of which express a two-fold dis- tinction where volitionality seems to interact with evidentiality. 8. Closing words
In this section, I will address some issues that might be of interest for further research. As the papers of this thesis contain rather detailed assessments of knowledge gaps that remain in the research on TAE in MM, Khorchin and Khalkha, I will restrict myself to a partial synthesis of the two independent papers on Khalkha, some additional comments on areal developments in Greater Mongolia, and a short note on the classification of CMc.
8.1 Evidentiality in Khalkha: a synthesis
Evidentiality in Khalkha turns out to be of some interest from a theoretical viewpoint. The language has two distinct systems, one for the present and one for the past. Both systems contain prominent neutral elements, so that neither can be accounted for by an explanation that restricts the overarching epistemological notion, be it called “evidentiality” or differently, to “source of information.” For past evidentiality in Khalkha, the first notion is the time of acquisition of information, and while this notion can be connected to surprise, attempts to draw too close of a link between it and to grammaticalized mirativity are suspect (see Hill 2012). Similarly, non-assimilated non-sensory evidentiality in Khalkha doesn’t exhibit any clear resemblance to “mediativity” (Lazard 1999) or “indirectivity” (Johanson 2006) if the latter is understood as a grammaticalized reference to cognitive or perceptive processes as the basis for asserting a particular state of affairs - which would thus relativize an assertion by mentioning that the information was acquired in some way. In contrast to the above, the difference between newly acquired vs. assimilated
40 information might at any rate merit closer attention, if not as an integral part of evidentiality, then at least as part of the heuristic notions considered dur- ing language description. Next to Khalkha and Tibetan (cf. Tournadre 2008), it also plays an important role in Kashaya, a Pomo language spoken in North America. Here, the distinction between different forms of direct and indirect sensory evidence available to the speaker is only drawn for non-assimilated information (see Oswalt 1986, esp. 30, 33). Subordinate to the time of acquisition are the canonical evidential notions “source of information” (speaker vs. known or unknown other person) and “mode of access” (sensory vs. inferential, plus subcategories): the speaker’s direct sensory perception of the event contrasts with inference and, arguably, information from other people. At least for the sample of spoken materials investigated, extended modal notions such as “confirmativity” (Friedman 1981) don’t seem to obtain in “spoken” Khalkha: the only use of a narrowly evidential marker in the corpus that diverges from a firm evidential notion is the use of -l=uu for recollection of events that need not be based on earlier direct sensory perception. It wouldn’t be surprising, though, if written lan- guage usages, both in novels and newspapers, diverged towards an affirma- tive respective distancing stance for -laa resp. -jee . Languages like Khalkha and Udihe also point to an interdependence of imperfective aspect and the expression of evidentiality. Knowledge about events that are unambiguously ongoing is likely to be firsthand, so Udihe uses a dedicated Firsthand Progressive, while reference to non-firsthand events that are ongoing can only be made by an aspectually unspecific form. On the other hand, direct synchronic access to an event that is ongoing is much more straightforward than assessing such evidence for habitual or ge- neric events. This justifies evidentiality as an obligatory category for ongo- ing events and persisting resultant states. While the synchronic access to evidence seems to indicate that the event has not yet been assimilated by the speaker, the synchronic absence of evidence is not as clearly determined in this respect. While the prototypical progressive event has not yet been assim- ilated, progressive first person forms verbs of cognition do refer to infor- mation about which the assertor has already made up her mind. Conversely, evidential marking is non-obligatory and, indeed, strongly dispreferred (in a ratio of 1:60 in my corpus) with habitual events, because they already tend to be established as valid and, thus, favor a factual, source-neutral stance. However, if such marking does occur, it seems that direct and indirect syn- chronic evidence is used to draw conclusions about a habitual event. In con- trast to progressive, continuative, perfect and prospective predications, the phase from which this evidence is taken would then be detached from any actual phase of the event, which should not facilitate processing. It is of interest that both the past and present system make a very similar primary distinction. On the one hand, they code synchronically available information (-n,-ѦD ,-WѻH ) or, in the case of the past events, information that
41 is still vividly accessible and cognitively activated in the mind. On the other hand, they code information already known to the speaker (-Vъ ի ƾ, unmarked) which might already be reified in her mind as a fact. Only for indirect evi- dence (-a), both assimilated and non-assimilated information is common, and finer distinctions are made through conventionalization of particular forms. One might speculate that this overall distinction was first developed on the present tense level. The past tense system might well have started out as a simple opposition between events taken to be factual and unproblematic and events about which the addressee should be enabled to judge herself on the basis of the speaker’s source of information / mood of access. A reinter- pretation into the present contrast might then have taken place subsequently. However, as I undertook my investigation into MM before I investigated Khalkha and even in that investigation mostly relied on Street (2008, 2009), I cannot tell whether a nuance such as recent acquisition of information might be traced back to quotes of direct speech in MM. While they are not part of the evidentiality system, the non-interrogative uses of -w (see 4.4), the non-habitual use of -Wъ ի J (see 4.3), and the idiosyn- cratic suffix plus particle combination -QѦH (briefly treated in Brosig 2012) are devices usable in contexts in which a narrator presents surprising, inter- esting events. Past -w seems to focus on the surprise that the speaker felt herself at the time of the event, and preventive -w focuses at an undesirable event that the addressee isn’t expecting. Non-habitual -Wъ ի J might mainly focus on a conflict between the speaker’s expectations given a normal course of events and the actual course of events. -QѦH might mostly be about catch- ing the addressee’s attention by suggesting that some event is unusual. Inas- much as Eerdunmengke (2009: 3.6.1.1) describes this form as expressing recollection in the closely related Ordos dialect, one may speculate that the recollection of an interesting detail has been conventionalized here. None of these markers mark “information which is new or unexpected to the speaker” (DeLancey 2001: 370): -w either concerns information that WAS unexpected or information that is unexpected (and, in the speaker’s opinion, undesirable) for the ADDRESSEE. The information marked by -Wъ ի J is usually not new, and while unexpectedness applies, it can be restricted to the past in narrative contexts. With -QѦH , again, it is mainly the addressee who is at issue. If one is willing to define mirativity as “a linguistic category that characterizes a proposition as newsworthy, unexpected, or surprising” either to the speaker or addressee (Hengeveldt & Olbertz 2012: 488), then ALL of these uses are mirative. I think it is crucial to consider, though, that all three devices are peripheral usages of rather old morphemes, and, in the case of -w, even the last surviving ones. In all three cases, the usages are so specialized that their text frequency is low. In the case of -w, one might speak of degrammaticali- zation (see Norde 2010) in the senses of resemantization of the bleached perfective past -bA and deparadigmaticalization. Semantically, if -Vъ ի ƾ can indeed be presumed to express an evidentially neutral, but an already assimi-
42 lated claim to the validity of a statement, one might conceive that a MM suffix -bA that was presumably neutral towards time of acquisition of infor- mation retained as its last domain the expression of evidentially unproblem- atic, but still not fully assimilated information.
8.2 Perspectives for areal studies on TAE in Greater Mongolia
In sections 6 and 7, I have tried to put the development of evidentiality in CMc from section 5 into an areal context. For Khorchin, it was found that the decline of evidential and aspectual constructions can be understood in the context of contact with Manchu and Mandarin, languages which lack gram- maticalized evidentiality. For Khalkha, the situation turned out to be more complicated: The past tense system, as inherited from MM, resembles the Turkic languages of Central Asia in that evidentiality is expressed by three markers that only seem to express two evidentiality values (in the canonical sense of the word). Detailed research into the function of such systems (such as Street 2009, Straughn 2011) is rather rare, while they seem to absent from the larger discussion of evidentiality. 40 On the other hand, the marking of present tense evidentiality differs substantially from Turkic and SMc, but shows recognizable parallels to eastern Tungusic languages and Amdo Ti- betan. Only the latter could directly have influenced Khalkha via religious contacts, though. If contact-induced change was due to Amdo Tibetan, the result seems rather notable, as I am not aware of languages other than Khal- kha and probably Buryat that have two relatively independent, tense- dependent, fully grammaticalized evidentiality systems (e.g., in comparison to non-tensed or tense-independent systems such as in Quechua [Faller 2002] or the Greater Tibetan area [e.g., Sun 1993, Tournadre and Dorje 2003, Åkerman 2012]). Another issue is the overall complexity of tense-aspect systems. In de- scriptions of neighboring languages, I didn’t find anything comparable to Khalkha when it comes to the complexity of TA markers. This might be due to underdescription or, more likely, the restricted number of sources I was able to use. 41 Conversely, the TA systems of neighboring varieties might actually be smaller. Then, Khalkha is quite reminiscent of Nakh-Dagestanian
40 Funnily, when Aikhenvald (2014: 9, 42, passim ) discusses some characteristics of the evi- dential system of Kalmyk, she takes it to be a system with two basic markers because she exclusively relies on Skribnik & Seesing (2014). The latter don’t mention -w but probably felt they didn’t have to because they didn’t intend to describe evidentiality in Kalmyk within its aspecto-temporal context and conceived of -w as evidentially neutral. 41 In particular, the collection of papers edited by Yakup et al. (2013) contains a number of important descriptions, many of which indicate additional problems and require the reader to follow up on cited references that are not easily accessible. As there is always the additional language barrier, I had to concede that I had obtained this book too late to make responsible use of it for this dissertation.
43 languages such as Chechen (as described by Molochieva 2010) that appar- ently show a similar ability to build up aspectually complex forms. For in- stance, it is possible in Chechen to mark an event as in progress at habitually reoccurring occasions merely through morphology and the use of one copu- la, as it is in Khalkha. If these similarities are due to one sprachbund, the Turkic languages in the west of Mongolia are the most likely link, though languages in the northwest such as Tuvan and Ket should probably not be excluded a priori from investigation either. As far as mature linguistic phenomena such as irregular verb stems are concerned, their lack in Mongolic, Turkic, Mandarin and by and large Man- chu forms a consistent area, while languages in geographically less accessi- ble areas such as Tibetan, the other Tungusic languages or Nakh- Dagestanian languages employ irregular stems in verb inflection. Another topic which merits areal investigation is the existence of tempo- rally discontinuous markers such as -Ѧъ that, depending on the overall con- text, can refer to both past and future, but not to the present (also cf. Nelson et al. 1998). While I argued in section 6 that the influence of Mandarin le might be at work here, this would mean that a phenomenon spread from Khorchin to Khalkha and, in the latter, even extended to the other past mark- ers. This is not entirely impossible, but similar markers in other languages bordering Khalkha might provide a more convincing areal context. It is pos- sible that a detailed analysis of Classical Mongolian sources could reveal the precise grammaticalization paths.
8.3 A short note on the classification of Central Mongolic
The classification still used in the “Ethnologue” (Lewis et al. 2014) is that of Oirat-Khalkha containing Khalkha-Buryat, which in turn consists of Khalkha Mongolian and Peripheral Mongolian. The latter is more or less defined as Mongolian spoken in China. The logic behind this classification is the exist- ence of Kalmykia, Buryatia, and Manchu-ruled Mongolia as political entities of the 19 th century and the more recent independence of Outer Mongolia as the Mongolian state in the early 20 th century. There is also a strong tendency to (more often implicitly than explicitly) take the language produced in writ- ing by speakers of different dialects as a coherent entity, as is done systemat- ically by non-dialectologists in Inner Mongolia and the Mongolian state. Chuluu (1998) from Eastern Inner Mongolia, for instance, considered his own idiolect as representative of Mongolian as spoken in Inner Mongolia, which is hardly appropriate even if we take the common Inner Mongolian standard (e.g., Qaserdeni et al. 2006), which is based on the Chakhar dialect 6HFHQEDܵDWXU %DW]D\DD LQ WXUQ FRPSDUHG WKH VWDQGDUGV RI the four largest political entities, also taking “Inner Mongolian” ( övörlögch ayalguu ) as an entity of description instead of calling it “Chakhar.” Howev-
44 er, given that the title of her book is “A phonological comparison of the ma- jor Mongolian dialects,” her point seems to lie in the justification of existing territorial divisions on a linguistic basis rather than finding out anything about actual Mongolian dialect groups. The problem is not a lack of data- based classifications (e.g., Cinggeltei 1957-58, Luvsanvandan 1959, Janhunen 2006) but that these are conveniently ignored. One of the points that I hopefully was able to show in this dissertation is that Khalkha- Chakhar, Oirat and Khorchin-Kharchin of CMc are all linguistically distinct enough to merit separate investigation. While I didn’t discuss the evidence, both the (rather brief) analysis RI6HFHQEDܵDWXU DQGP\RZQLPSUH s- sion from interacting with Chakhar speakers indicate a TAE system that very closely resembles Khalkha and, in particular, its southern dialects. My point is that in not considering closely related varieties spoken in other countries, scientists miss applicable research results, and in taking countrilects as ob- jects of description, grammarians must either become selective without a proper scientific basis or describe very fuzzy entities. You can either be po- litically overcorrect, or a scientist. Bibliography
Aikhenvald, Alexandra. 2004. Evidentiality . Oxford: Oxford University Press. Aikhenvald, Alexandra. 2014. The grammar of knowledge: a cross-linguistic view of evidentials and the expression of information source. In: Alexandra Ai- khenvald & R. Dixon (eds.), The grammar of knowledge . Oxford: Oxford Uni- versity Press: 1-51. (uncorrected proof used) Al-.Ɨã÷DUƯ0DতPRXG &RPSHQGLXPRIWKH7XUNLFGLDOHFWV 'ƯZƗQ/XћƗWDW - Turk) . Edited and translated by Robert Dankoff with James Kelly. Cambridge: Harvard University Press. Åkerman, Vesa. 2012. Inflection of finite verbs in Mongghul. SIL Electronic Work- ing Papers 2012–003 : 1-60. Atwood, Christopher. 2013. From Bagisba to Pagba: Mongolian systems for writing Tibetan. Presentation at the XIII conference of the International Association for Tibetan Studies, Ulaanbaatar, July 21-27. Barkmann, Udo. 1999. Geschichte der Mongolei . Bonn: Bouvier. Batzayaa, Gerelt-Odiin. 2008. Mongol helnii nutgiin ayalguunuudiin avianii har’cuulal . Ulaanbaatar: Mongol ulsiin bolovsroliin ih surguul’. %D\DQFRܵWX 4RUþLQ DPDQ D\DOћXQ -u sudulul . Kökeqota: Öbür mong ܵul-un \HNHVXUܵDܵXOL -yin keblel-ün qoriy-a. %D\DUPHQG 'XPGDGXMDܵXQ -XPRQJܵXONHOHQ -ü temdeg üile üge-yin kelberi- nügüd ba tegün-ü ularil kögjil. gEUPRQJћXO -XQ\HNHVXUћDћXOL , 1986/1: 49-81. Benzing, Johannes. 1985. Kalmückische Grammatik zum Nachschlagen . Wiesbaden: Harrassowitz. Bertinetto, Pier Marco, & Alessandro Lenci. 2012. Habituality, pluriactionality, and imperfectivity. In: Robert Binnick (ed.), The Oxford handbook of tense and as- pect . Oxford: Oxford University Press: 852-80. Bese, Lajos. Remarks on Middle Mongolian conjugation. Acta Orientalia 32: 21-35.
45 Bese, Lajos. 1971. Problémák a mongol igeragozás történetében. Nyelvtudományi Közlemények LXXIII : 217-22. Binnick, Robert. 2012. The past tenses of the Mongolian verb . Leiden: Brill. Bläsing, Uwe. 1984. Die finit indikativischen Verbalformen im Kalmückischen . Stuttgart: Franz Steiner. Botne, Robert. 2012. Remoteness distinctions. In: Robert Binnick (ed.), The Oxford handbook of tense and aspect . Oxford: Oxford University Press: 536-63. Breu, Walter. 1988. Resultativität, Perfekt und die Gliederung der Aspektdimensio- nen. In: R. Raecke (ed), Slavische Linguistik 1987. Referate des XIII. Konstan- zer Slavistischen Arbeitstreffens . München: Otto Sagner: 42-74. Breu, Walter. 1996. Komponentenmodell der Interaktion von Lexik und Aspekt. In: W. Girke (ed.), Slavistische Linguistik 1995 . München: Sagner: 37-74. Breu, Walter. 2005. Verbalaspekt und Sprachkontakt: Ein Vergleich der Systeme zweier slavischer Minderheitensprachen (SWR/MSL). In: Sebastian Kempgen (ed.) Slavistische Linguistik 2003 . München: Sagner: 37-95. Brosig, Benjamin. 2009. Die satzfiniten Aspektmarker des modernen Khalkha- Mongolischen. Bonn: Universität Bonn. Unpublished magister thesis. Brosig, Benjamin. 2012. “Bilee” sul ügiin utga, hereglee (The meaning and usage of the particle “bilee”). Hel zohiol sudlal, V (37) : 10-18. Buder, Anja. 1989. Aspekto-temporale Kategorien im Jakutischen . Wiesbaden: Har- rassowitz. Byambasan, P., C. Önörbayan, B. Pürev-Ochir, J. Sanjaa, & C. Janchivdorj. 1987. Orchin cagiin mongol helnii ügzüin baiguulalt (The structure of [verbal] mor- phology of contemporary Mongolian). Ulaanbaatar: Shinjleh uhaanii akadyemi. Bybee, Joan, Revere Perkins, & William Pagliuca. 1994. The evolution of grammar . Chicago: University of Chicago Press. >&DܵDQTDGD@ &KiJƗQKƗGi 0ƝQJJ·\· NƝČUVKqQ W·\· \iQMLnj %ČLMƯQJ Shèhuì NƝ[XpZpQ[LjQFKnjEӽQVKq Chuluu, Üjiyediin. 1994. Introduction, grammar, and sample sentences for Jegün Yogur . Sino-Platonic papers 54. Philadelphia: University of Pennsylvania. Chuluu, Üjiyediin. 1998. Studies on Mongolian verb morphology. Toronto: Univer- sity of Toronto. Unpublished doctoral thesis. Cinggeltei. 1957-1958. Dumdadu ulus-WHNL PRQJܵXO W|UO -ün kele-QXܵXG ED PRQJܵXO NHOHQ - D\DOܵX -QXܵXG -un yeröngkei bayidal. 0RQJћXO -un kele bicig 1957.11-12, 1958.1-7. &LQJJHOWHL 0RQJܵXO NHOHQ -ü tusalaqu üile üge-yin tuqai. gEU PRQJћXO -un \HNHVXUћDћXOL : 1-103. Comrie, Bernhard. 1985. Tense . Cambridge: Cambridge University Press. Creissels, Denis. 2008. Language documentation and verb inflection typology: the case of Northern Akhvakh (Nakh-Daghestanian). Paper presented at Chronos 9, 2-4 of October, Paris. Croft, William. 2012. Verbs . Oxford: Oxford University Press. Dahl, Östen. 1984. Temporal distance: remoteness distinctions in tense-aspect sys- tems. In: Brian Butterworth, Bernard Comrie, & Östen Dahl (eds.), Explana- tions for language universals . Berlin: Mouton: 105-22. Dahl, Östen. 1985. Tense and aspect systems . Oxford: Blackwell. de Haan, Ferdinand. 2001. The place of inference within the evidential system. In- ternational Journal of American Linguistics 67 : 193–219. DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33: 369-82. de Rachewiltz, Igor. 1999. Some reflections on so-called Written Mongolian. In: Helmut Eimer, Michael Hahn, Maria Schetelich & Peter Wyzlic (eds.), Studia
46 Tibetica et Mongolica - Festschrift Manfred Taube. Swisttal-Odendorf: Indica et Tibetica Verlag: 235-46. Dowty, David. 1979. Word meaning and Montague grammar . Dordrecht: Reidel. Dugarova, Galina Sonomovna. 1991. Glagol’nyj vid v sovremennom mongol’skom jazyke . Novosibirsk: Nauka (Sibirskoe otdelenie). Ebert, Karen. 1999. Degrees of focality in Kalmyk imperfectives. In: Werner Abra- ham & Leonid Kurilov (eds.), Tense-aspect, transitivity and causativity . Ams- terdam: John Benjamins: 323-39. Eerdunmengke. 2009. Zur phonetischen und grammatischen Entwicklung der mon- golischen Ordos-Sprache. Unpublished manuscript. Erdal, Marcel. 2004. A grammar of Old Turkic . Leiden: Brill. Faehndrich, Burgel. 2007. Sketch grammar of the Karlong variety of Mongghul, and dialectal survey of Mongghul. University of Hawai’i. Unpublished dissertation. Faller, Martina. 2002. Semantics and pragmatics of evidentials in Cuzco Quechua. Stanford: Stanford University. Unpublished dissertation. Fried, Robert Wayne. 2010. A grammar of Bao’an Tu, a Mongolic language of northwest China. University of Buffalo. Unpublished dissertation. Friedman, Victor. 1981. Admirativity and confirmativity. Zeitschrift für Balkanolo- gie 17.1: 12-28. Goto, K. V. 2009. Sistema finitnych form prošedšego vremeni v kalmyckom jazyke [The system of finite past tense forms in Kalmyk]. Acta linguistica petropolita- na V.2: 124-59. Gorelova, Liliya M. 2002. Manchu grammar . Leiden: Brill. Haller, Felix. 2004. Dialekt und Erzählungen von Themchen . Bonn: VGH Wissen- schaftsverlag. Hengeveldt, Kees, & Hella Olbertz. 2012. Didn’t you know? Mirativity does exist! Linguistic Typology 16.3: 487-503. Henne, Henry, Ole Bjørn Rongen, & Lars Jul Hansen. 1977. A handbook on Chinese language structure . Oslo: Universitetsforlaget. Higuchi, Koichi. 1994. Tibetanisms in the Mongolian Buddhist literature. In: Hajime Kitamura, Tatsuo Nishida, & Yasuhiko Nagano (eds.), Current issues in Sino- Tibetan linguistics . Osaka: Organizing Committee the 26th International Con- ference on Sino-Tibetan Languages and Linguistic: 1073-79. Hill, Nathan. 2012. “Mirativity” does not exist - তGXJLQ³/KDVD´7LEHWDQDQGRWKHU suspects. Linguistic Typology 16/3: 389-433. Hsiao, Suying. 2013. The grammatical temporal system from Middle Mongolian to Modern Mongolian. Language and Linguistics 14/6: 1075-103. Janhunen, Juha. Manchuria - an ethnic history . Helsinki: Suomalais-Ugrilainen Seura. Janhunen, Juha. 2003. Proto-Mongolic. In: Juha Janhunen (ed.), The Mongolic lan- guages . London: Routledge: 1-29. Janhunen, Juha. 2005. Khamnigan Mongol . München: Lincom. Janhunen, Juha. 2006. Mongolic languages. In: K. Brown (ed.), The encyclopedia of language & linguistics . Amsterdam: Elsevier: 231-34. Janhunen, Juha. 2012. Khitan - Understanding the language behind the scripts. SCRIPTA , Vol. 4: 107-32. Janhunen, Juha, Marja Peltomaa, Erika Sandman, & Xiawu Dongzhou. 2008. Wutun . München: Lincom Europa. Johanson, Lars. 1999. Grenzbezogenheit in Aspekt und Lexik am Beispiel türkischer Postverbialkonstruktionen. In: Walter Breu (ed.), Probleme der Interaktion von Lexik und Aspekt (ILA). Tübingen: Max Niemeyer: 129-39.
47 Johanson, Lars. 2000. Viewpoint operators in European languages. In: Östen Dahl (ed.), Tense and aspect in the languages of Europe . Berlin & New York: Mou- ton de Gruyter: 27-187. Johanson, Lars. 2003. Evidentiality in Turkic. In: A. Y. Aikhenvald & R. M. W. Dixon (eds.), Studies in evidentiality. Amsterdam: Benjamins: 73-290. Johanson, Lars. 2006. Indirective sentence types. Turkic Languages 10: 73-89. Kane, Daniel. 2009. The Kitan language and script . Leiden: Brill. Kim, Gi-Song. 1995. Orchin cagiin mongol helnii üil ügiin engiin cag, tüünii togtolcoo (The simple tenses of contemporary Mongolian and their system). Doctoral thesis (published 1998), Mongol ulsiin ündesnii ih surguul’. Kiryu, Kazuyuki. 1999. Conceptualization and aspect in some Asian languages. In: Werner Abraham and Leonid I. Kulikov (eds.), Tense-aspect, transitivity and causativity . Amsterdam: Benjamins: 43-62. Klein, Wolfgang. 1994. Time in language . New York: Routledge. Ko, Seongyeon. 2011. Vowel contrast and vowel harmony shift in the Mongolic languages. Language Research 47.1: 23-43. Kullmann, Rita, & Dandii-Yadamiin Tserenpil. 1996. Mongolian grammar . Ulaanbaatar: Admon. .UHOEDܵDWXUDQG8UDQFLPHJ Qorcin-XMDQJDћDOL .|NHTRWD gEUPRQJܵXO - un arad-un keblel-ün qoriy-a. Lazard, Gilbert. 1999. Mirativity, evidentiality, mediativity, or other? Linguistic Typology 3.1 : 91-110. Lewis, M. Paul, Gary F. Simons, & Charles D. Fennig (eds.). 2014. Ethnologue. 17 th edition. Dallas, Texas: SIL International. Li, Charles, Sandra Thompson, & R. McMillan Thompson. 1982. The discourse motivation for the perfect aspect: the Mandarin particle le . In: Paul Hopper (ed.), Tense-aspect: between semantics & pragmatics . Amsterdam: Benjamins. Li, Mei. 1999. Negation in Chinese. Manchester: University of Manchester. Un- published doctoral thesis. Luvsanvandan, Sh. 1959. Mongol hel ayalguunii uchir. Mongoliin sudlal , 1. Malchukov, Andrej. 2000. Perfect, evidentiality and related categories in Tungusic languages. In: Bo Utas and Lars Johanson (eds.), Evidentiality in Turkic, Iranian and neighbouring languages . Berlin: Mouton de Gruyter: 441-70. 0DWVXRND <njWD ᯇᒸ㞝ኴ ). 2008. Mongorugo-no asupekuto-ni kan-VXUX NHQN\nj (A study of aspect in Mongolian) )XNXRND .\njVKnj 8QLYHUVLW\ 8QSXEOLVKHG doctoral dissertation. Michael, Lev. 2006. “The moral implications of evidentiality in Nanti society: Epis- temic distance as a pragmatic metaphor for moral responsibility.” Proceedings of the Thirteenth Symposium about Language and Society - Austin, Vol. 13 . Mijiddorj, G. 1976. Mongol, manj bichgiin helnii har’caa . Ulaanbaatar: Shinjleh uhaanii akademi. Molochieva, Zarina. 2010. Tense, aspect, and mood in Chechen. Leipzig: Universi- tät Leipzig. Unpublished doctoral thesis. Mönggüngerel. 1998. 1DLPDQ DPDQ D\DOћX .|NHTRWD gEU PRQJܵXO -un yeke VXUܵDܵXOL -yin keblel-ün qoriya. Mönh-Amgalan, Yumjiriin. 1998. Orchin cagiin mongol xelnii baimjiin ai (Modality in contemporary Mongolian). Ulaanbaatar: Mongol ulsiin ih surguul’. Mushin, Ilana. 2001. Evidentiality and epistemological stance - narrative retelling. Philadelphia: Benjamins. Nelson, Diane, T. Davidson, Narantuya, Narmandah, & Judith Nordby. 1998. Four past tense morphemes in Khalkh Mongol. Leeds Working Papers in Linguistics and Phonetics 6 : 115-31.
48 Nikolaeva, Irina, & Maria Tolskaya. 2001. A grammar of Udihe . Berlin: Mouton de Gruyter. Norde, Muriel. 2010. Degrammaticalization: three common controversies. In: Kate- rina Stathi, Elke Gehweiler, & Ekkehard König (eds.), Grammaticalization . Amsterdam: John Benjamins: 123-50. Nugteren, Hans. 2003. Shira Yugur. In: Juha Janhunen (ed.), The Mongolic lan- guages . London: Routledge: 265-85. Nugteren, Hans. 2011. Mongolic phonology and the Qinghai-Gansu languages. Utrecht: LOT. Orulamjab. 2013. 0RQJܵXONHO en-WRUܵXGD\DOܵXQ -u üile üge-yin sudulul (Verbs in Torgut Mongolian). Beijing: Central University of Nationalities. Unpublished master thesis, pre-final version, 2013-05-26. Oswalt, Robert. 1986. The evidential system of Kashaya. In: Wallace Chafe & Jo- hanna Nichols (ed.), Evidentiality . Norwood: Ablex: 29-45. Plungian, Vladimir. 2011. Types of verbal evidentiality marking: an overview. In: Gabriele Diewald & Elena Smirnova (eds.), Linguistic realization of evidentiali- ty in European languages . Berlin: De Gruyter Mouton: 15-58. 3RSSH 1LNRODM 1LNRODHYLþ Introduction to Mongolian comparative studies . Helsinki: Suomalais-ugrilainen seura. 'RUML Mong ћul DL 7L\HQVL\XXۥ4DVHUGHQL 6HFHQ %XX 0DQOL\DQJ 6DQJ yariyan-u kele jüi . Kökeqota: Öbür mong ܵul-un arad-un keblel-ün qoriy-a. Rentzsch, Julian. 2012. Scope hierarchies and modality in Manchu. Studies in Lan- guage 36/4: 848-81. Rybatzki, Volker. 2003. Middle Mongol. In: Juha Janhunen (ed.), The Mongolic languages . London: Routledge: 47-82. Sasse, Hans Jürgen. 2002. Recent activity in the theory of aspect. Linguistic Typolo- gy 6: 199-271. Sasse, Hans-Jürgen. 2006. Aspect and aktionsart. In: K. Brown (ed.), The encyclo- pedia of language & linguistics . Amsterdam: Elsevier: 535-38. >6HFHQEDܵDWXU@ 6HFKHQEDDWDU %RUMLJLQ The Chakhar dialect of Mongol: a morphological description . Helsinki: Finno-Ugrian society. Schönig, Claus. 2003. Turko-Mongolic relations. In: Juha Janhunen (ed.), The Mon- golic languages . London: Routledge: 403-19. Secen, Ca, 1987. Dumdadu ja ܵXQ -X PRQJܵJXONHOHQ -ü tusalaqu üile üge-nügüd ba tegün-ü ularil kögjil. .|NHTRWD gEUPRQJܵXO -XQ\HNHVXUܵDܵXOL8QSXEOLVKHG doctoral [?, sudulun surugci-yin] thesis. 6HFHQEDܵDWXU 4DVJHUHO 7X\Dܵ -a [Tuyaa], Bu. Jirannige, & U Ying je [Wú, <ƯQJ]Kp ྻⱥဴ ]. 2005. 0RQJћXO NHOHQ - QXWXћ -XQ D\DOћXQ -X VLQЋLOHO -ün uduridqal [A guide to the regional dialects of Mongolian]. Kökeqota: Öbür PRQJܵXO -un arad-un keblel-ün qoriy-a. Serjee, Jambaldorjiin. 2010. Orchin cagiin mongol helnii onooson neriin sudalgaa. Ulaanbaatar: Mongol ulsiin ih surguul’. Shiotani, Shigeki ( ሷ㇂ⱱᶞ ). 2005. Mongol töröl helnüüd deh “tuhain üiliig üildeh dur sonirholtoi, zanshiltai” gesen utga sanaag ilerhiileh üilees ner büteeh daga- vruudiin tuhai. Osaka gaikokugo d DLJDNXURQVKnj 31: 115-44. Skribnik, Elena, & Olga Seesing. 2012. Kalmückische analytische Verbalkon- struktionen mit dem Perfektpartizip: Evidentialis und Mirativ. In: Tiina Hyytiäi- nen, Lotta Jalava, Janne Saarikivi, & Erika Sandman (eds.), Per Urales ad Ori- entem - Festskrift tillägnad Juha Janhunen på hans sextioårsdag den 12 febru- ari 2012 . Helsinki: 409-24.
49 Skribnik, Elena, & Olga Seesing. 2014. Evidentiality in Kalmyk. In: Alexandra Aikhenvald & R. Dixon (eds.), The grammar of knowledge . Oxford: Oxford University Press: 148-70. Slater, Keith. 2003. A grammar of Mangghuer . London: Routledge Curzon. Sneath, David. 2000. Changing Inner Mongolia – Pastoral Mongolian society and the Chinese state . Oxford: Oxford University Press. [Sodu] Šuvangšan. 2013. Qaracin-qorc LQ D\DOܵX ED WDUL\DODTX VR\XO In: [Bür- insayin] Borujigin Burensain (ed.), 1DLPRQJRUX WǀIX FKLLNL QL RNHUX WHLMǀ WR QǀNǀNDQRVRNXVHNL . Nagoya: Nagoya University: 65-93. >6RGXEDܵDWXU@ &iRGjREƗWqČU .ƗOTuQ PpQJJ·\· \iQML (A grammar of Kharchin Mongolian) %ČLMƯQJ 0tQ]~FKnjEӽQVKqFKnjEӽQIƗ[tQJ Song, Jae-mog. 1997. Tense, aspect and modality in Khalkha Mongolian. Un- published doctoral thesis, University of London, SOAS. Straughn, Christopher. 2011. Evidentiality in Uzbek and Kazakh. Chicago: Univer- sity of Chicago. Unpublished dissertation. Street, John. 1986. The particle ele in Early Middle Mongolian. Journal de la Société FinnoOugrienne 80: 195-258. Street, John. 2008. Middle Mongolian past-tense -BA in the Secret History // Jour- nal of the American Oriental Society 128/3: 399-422. Street, John. 2009. On the three past tense endings of early Middle Mongolian. Ural-Altaische Jahrbücher 23 : 126-59. Street, John. 2013. On quotation in Middle Mongolian: the verb ke(m)e-'to say.' Bloomington: The Mongolia Society. >6nj<ӿZpQ@6X/LO\, -wen ( ⸽௨ᩥ ). 2004. Subjectification and the use of the com- plementizer SHUO. Concentric 30.1: 19-40. Sun, Jackson. 1993. Evidentials in Amdo Tibetan. Bulletin of the Institute of History and Philology , Academia Sinica 63 : 945-1001. Svantesson, Jan-Olof, Anna Tsendina, Anastasia Karlsson, & Vivan Franzén. 2005. The phonology of Mongolian . New York: Oxford University Press. Talmy, Leonard. n.a. Introspection as a methodology in linguistics . http://linguistics.buffalo.edu/people/faculty/talmy/talmyweb/Handouts/introspec tion2.pdf . Tournadre, Nicolas, & Sangda Dorje. 2003. Manual of Standard Tibetan . Translated by Charles Ramble. Ithaca, New York: Snow Lion Publications. Tournadre, Nicolas. 2008. Arguments against the concept of ‘conjunct’/‘disjunct’ in Tibetan. In: Brigitte Huber, Marianne Volkart, & Paul Widmer (eds.), Chomo- langma, Demawend und Kasbek, Festschrift für Roland Bielmeier. Halle: IITBS: 281-308. Tsai, Sheng-Luen. 1983. Chinese settlement of Mongolian lands: Manchu policy in Inner Mongolia: A case study of Chinese migration in Jerim league. Provo, Utah: Brigham Young University. Unpublished doctoral thesis. Tatevosov, Sergei. 2001. From resultatives to evidentials - Multiple uses of the Per- fect in Nakh-Daghestanian languages. Journal of Pragmatics 33 : 443-64. Tatevosov, Sergei. 2002. The parameter of actionality. Linguistic Typology 6 : 317- 401. Tatevosov, Sergei. 2008. Subevental structure and non-culmination. Empirical Is- sues in Syntax and Semantics 7 : 393–422. 7RܵWDPED\DU / .HOHQ -ü üjelte kiged qubi kümün-ü kelen-ü kereglegen-ü qogurunduki qaricag-a. Monggul-un sudulul 2007/3 : 39-49. 7RܵWDPED\DU/ 0RQJћXONHOHQ -ü kele jüijil kiged jokicamji . Šenyang: Liyu- uning-un ündüsüten-ü keblel-ün qoriy-a.
50 [Tömörtogoo, Domiin] Tumurtogoo, D. 2006. Mongolian monuments in Uighur- Mongolian script (XIII-XVI centuries) . Taipei: Academia Sinica. von Möllendorff, P. G. 1892. A Manchu grammar with analysed texts . Shanghai: American Presbyterian Mission Press. Vovin, Alexander. 2007. Once again on the 7DEJDþ ODQJXDJH Mongolian Studies XXIX: 191-206. Wáng, <~IƗQJ ⋤ⴝⰾ ), Aya Katz, & Chih-Hua Chen. 2003. Thinking as saying: shuo (‘say’) in Taiwan Mandarin conversation and BBS talk. Language Sci- ences 25: 457-88. Weiers, Michael. 1972. Die Sprache der Moghol der Provinz Herat in Afghanistan . Opladen: Westdeutscher Verlag. Xiao, Richard, & Tony McEnery. 2004. Aspect in Mandarin Chinese . Amsterdam: Benjamins. 51